Before the FEDERAL COMMUNICATIONS COMMISSION Washington, DC ) ) ) ) ) REPLY COMMENTS OF T-MOBILE USA, INC.

Transcription

1 Before the FEDERAL COMMUNICATIONS COMMISSION Washington, DC In the Matter of Expanding the Economic and Innovation Opportunities of Spectrum Through Incentive Auctions ) ) ) ) ) WT Docket No REPLY COMMENTS OF T-MOBILE USA, INC. Ari Fitzgerald Trey Hanbury Phillip Berenbroick AJ Burton Hogan Lovells US LLP 555 Thirteenth Street, NW Washington, DC (202) Attorneys for T-Mobile USA, Inc. Thomas Sugrue Kathleen O Brien Ham Steve Sharkey Christopher Wieczorek Joshua Roland Indra Chalk T-Mobile USA, Inc. 601 Pennsylvania Avenue, NW Washington, DC (202) March 12, 2013

2 TABLE OF CONTENTS Page I. Introduction... 1 II. The Proposed 600 MHz Band Plan... 2 A. Widespread Support Exists for a Number of Core Elements in the 600 MHz Band Plan Maximize Paired Spectrum Maximize Opportunities for Competition During and After the Auction Conduct a Timely Auction Establish a Fixed Amount of Downlink Spectrum Create Supplemental Downlink Spectrum for Asymmetric Pairing B. Commenters Agree that the Commission s Lead Band Plan Proposal Suffers from Shortcomings that Other Band Plans Can Address C. The 35x35 MHz Band Plan T-Mobile Has Proposed Harnesses the Benefits of the Commission s Lead Band Plan While Avoiding Many of the Risks Antenna Performance Duplex Filter Performance Harmonic Interference D. Band Plans that Provide Less than a Potential of 70 Megahertz of Paired Spectrum for Auction Will Decrease Auction Revenues and Harm Competition AT&T and Qualcomm s Band Plan Google and Microsoft A TDD Band Plan Has Serious Shortcomings, and Commenters Overwhelmingly Support an FDD Band Plan III. The Forward Auction A. The Record Supports the Adoption of a Spectrum Cap that Would Limit the Amount of Spectrum Below 1 GHz that a Single Licensee Can Hold Excessive Concentration in the U.S. Wireless Market Harms the Public Interest A Spectrum Cap Below 1 GHz Should Be Applied to Prevent Further Harm Caused by the Consolidation of Valuable Spectrum into the Hands of a Few Players The Commission Should Reject the Objections by Verizon and AT&T to the Spectrum Cap B. Random or Quasi-Random License Assignments Will Clear More Spectrum, Increase Efficiency, Promote Interoperability, and Raise More Revenue in the Forward Auction Creating Durable Incentives for Interoperability Resolving Concerns about Anonymous Bidding Making the Clearing Rule Easier to Satisfy ii

3 4. Raising More Revenue C. The Commission Should License the 600 MHz Spectrum By Major Economic Area D. The Commission Should Allow Package Bidding for Geographic Areas if It Adopts EAs, for Unpaired Blocks, and, if It Adopts Safeguards Against Abuse, for Paired Blocks E. Without Careful Oversight and Stiff Penalties for Abuse, Special Bidding Credits Will Harm the Auction and Damage Competition IV. The Reverse Auction A. Addressing the Most Important Reverse Auction Concerns As Soon As Possible Would Promote Broadcaster Participation and Accelerate the Incentive Auction B. Using Multiple Opening-Round Bids Promises a Simpler and Faster Auction that is More Likely to Satisfy the Clearing Rule than Other Alternatives C. Sequentially Alternating the Reverse and Forward Auctions Balances the Need for Simplicity and Speed Against the Desire to Allow Participants to Monitor and Respond to a Changing Auction Environment D. Offering Broadcasters Too Many Exit Options in the Reverse Auction Could Produce Excessive Uncertainty and Delay or Disrupt the Auction Process E. Addressing an Unsatisfied Closing Requirement F. While the Concept of Extended Rounds Has Great Promise, Modifications Are Needed and a Last-Call Solution May Offer a Simpler Solution to the Problem of a Clearing- Rule Shortfall G. Broadcasters Participating in the Incentive Auction Should Not Be Allowed to Revoke Their Bids During or After the Auction V. Relocation and Clearing A. The Repacking Process Should Seek to Maximize the Amount of Spectrum Reallocated for Flexible Use The Plain Text of the Spectrum Act Accords the Commission Substantial Flexibility in Repacking Broadcasters The Purpose of the Spectrum Act, Including Provisions Addressing Repacking of Broadcasters, is to Reallocate Spectrum for Broadband Deployment B. The Commission Has a Number of Options to Overcome the Challenges of Repacking Broadcasters C. Secondary Means Secondary D. Authorizing Unlicensed Use of 600 MHz Bands Prior to Commercial Deployment Will Decrease Auction Revenues and Delay Broadband Deployment VI. CONCLUSION iii

4 EXECUTIVE SUMMARY The 600 MHz incentive auction represents the country s best near-term opportunity to help satisfy consumers appetite for mobile broadband services. Careful planning of four features the 600 MHz band plan, the forward auction, the reverse auction, and the broadcast repacking promises to stimulate investment, promote competition, and accelerate mobile broadband deployment. The Band Plan Commenters were virtually unanimous in recommending that paired 600 MHz operations occur above Channel 37, not on either side of this channel. This configuration offers flexibility for variable levels of spectrum clearing and responds to the predominant view among commenters that the greatest value and largest opportunities for competitive entry and expansion lie in maximizing paired spectrum bands. Incorporating at least 35x35 MHz of spectrum in this configuration will not only offer competitive carriers an opportunity to acquire critical, highvalue low-frequency spectrum resources through competitive bidding, but also will allow up to three competitors each to acquire enough contiguous spectrum to operate at high levels of efficiency. So long as the Commission can license up to seven paired five-megahertz blocks, pairing the 84 megahertz above Channel 37 offers the greatest public benefits with the fewest and least extensive technical, economic and competitive deficiencies. The Forward Auction The incentive auction may fail if too few participants join the bidding and the new spectrum auction leads to less competition in the marketplace for wireless services rather than more competition, which would benefit consumers and the federal treasury. One of the strongest deterrents to widespread participation in the 600 MHz auction is the prospect that bidding will be iv

5 pointless if the nation s two largest carriers each of which has a market capitalization roughly ten times that of its next largest competitor are given an unfettered ability to acquire all of the spectrum offered. Most commenters, therefore, support imposing a cap on spectrum acquisitions, such as T-Mobile s proposal to limit bidders from acquiring more than one-third of the spectrum below 1 GHz. In a highly concentrated, capital intensive market such as mobile broadband, dominant firms have a strong economic interest in maintaining and increasing market power. Excluding rivals allows such firms to charge more for existing service and relieves competitive pressure to innovate and invest in new offerings. Absent some type of cap on acquiring critical spectrum resources, the two largest carriers will have an incentive to pay a premium during the competitive bidding process not because they can realize greater efficiencies using that spectrum, but rather because acquiring the spectrum will allow them to foreclose competitors from the market. during the 600 MHz auction. A pro-competitive cap on spectrum below 1 GHz must apply After-the-fact divestitures would prevent or delay competitive entry or expansion and allow the largest carriers to select spectrum for divestiture that potential rivals would not choose due to its limited synergies, paltry scale, high development costs, or other factors. Furthermore, if a small number of incumbent providers end up with control over large amounts of spectrum, after-the-fact divestitures will not provide timely or effective remedies for competitors or consumers. While a cap on spectrum acquisition during the auction is therefore necessary to ensure competition, it is not sufficient. Even if the two largest firms do not end up acquiring excessive amounts of spectrum within the 600 MHz band, they can raise rivals costs and impede consumer choice by developing devices useable only in those segments of the 600 MHz band that they control. A rule against anticompetitive equipment segregation that thwarts interoperability is v

6 therefore critical to the success of the band plan. A block-assignment policy that sufficiently diversifies holdings to prevent any one operator from exercising market power to foreclose rivals within the 600 MHz band from relying on spectrum blocks with the greatest scale will complement such a requirement. In assigning geographic area licenses under the lead band plan, the Commission should use Major Economic Area ( MEA ) licenses, rather than Economic Area ( EA ) licenses as its primary geographic area. With a broader geographic footprint than EA licenses, MEAs reduce the exposure risk that a carrier might win some, but not all of the licenses they need to provide an economical service in the 600 MHz band. The larger footprint of MEAs minimizes the need for complicated package bidding in the 600 MHz auction. EAs can be used for areas with sufficient additional spectrum clearing and provide an opportunity for smaller carriers to provide service as well. The Reverse Auction The Commission can best encourage widespread participation in the reverse auction by adopting certain procedures and safeguards designed to maximize the information available to participating broadcasters and to provide as much certainty as possible, as soon as possible. Specifically, the Commission should: Implement a sequential or staged approach to the reverse and forward auction, rather than a one-iteration reverse auction: A staged approach will provide broadcasters crucial information about the extent of buyer demand for broadcast spectrum licenses that will better enable them to value their licenses. Address the potential for unsatisfied closing requirements by adopting a take-itor-leave-it option at the end of the auction: A take-it-or-leave-it approach, which could be accomplished by one of several different mechanisms, would eliminate much of the uncertainty over how winning bidders will split the shortfall between their bids and the minimum price required by the clearing rule. vi

7 Limit broadcaster exit options to those that are provided by statute: Allowing broadcasters to submit other types of bids beyond those set forth in the Spectrum Act would unduly complicate the reverse auction, introducing uncertainty that would chill broadcaster participation in the auction rather than encourage it. Prohibit revocation of bids: The Commission should treat reverse auction bids as irrevocable and binding to ensure that broadcasters will bid truthfully in the reverse auction, thereby maximizing predictability and participation. The Repacking In devising a plan for the repacking process, the Commission should exercise its considerable discretion to implement measures designed to preserve the coverage area and population served by each relocated broadcaster to the extent reasonable, while promoting the auction s primary goal of reallocating spectrum for broadband deployment. In so doing, the Commission need not achieve identical coverage areas and service population. Instead, the Commission must only undertake reasonable efforts in the repacking process. The Commission can meet that obligation while simultaneously addressing the many logistical and structural challenges posed by the relocation of broadcasters. Indeed, certain measures can reduce uncertainty about the repacking, ensure that broadcasters are made whole for relocating, and expedite the deployment of the spectrum for wireless use all of which will encourage participation in the auction and maximize the amount of spectrum made available for broadband use. Specifically, the Commission should: (1) require broadcasters to provide inventories of equipment and facilities that will be affected by the repacking process and estimates of repacking costs; (2) gather information from television equipment manufacturers regarding the requirements, costs, and timeframes necessary for completing the broadcaster transition; (3) fix firm deadlines for various milestones that broadcasters must satisfy prior to receiving full payment for relinquishing spectrum rights or for reimbursement of relocation vii

8 costs; (4) implement a two-step reimbursement process whereby broadcasters receive an initial payment based on estimated relocation costs followed by a true up payment to reimburse additional reasonable costs that are incurred; and (5) encourage the Internal Revenue Service to permit reverse auction proceeds to be treated as an involuntary conversion such that the proceeds are deferred for income tax purposes. * * * With pro-competitive, pro-consumer policies in place, the broadcast incentive auction will stimulate investment, spur innovation, and benefit American consumers. viii

9 Before the FEDERAL COMMUNICATIONS COMMISSION Washington, DC In the Matter of Expanding the Economic and Innovation Opportunities of Spectrum Through Incentive Auctions ) ) ) ) ) WT Docket No REPLY COMMENTS OF T-MOBILE USA, INC. I. INTRODUCTION T-Mobile USA ( T-Mobile ) submits these reply comments in response to the Notice of Proposed Rulemaking ( NPRM ) issued by the Federal Communications Commission ( FCC ) in the above captioned proceeding. 1 With hundreds of comments filed in this proceeding, the Commission has started to assemble an extensive record to help resolve the four key features of the 600 MHz incentive auction: the 600 MHz band plan, the forward auction, the reverse auction, and the broadcast repacking. The comments reveal several key areas of consensus and numerous viable options for success. Many commenters endorse a band plan that maximizes paired spectrum, creates supplemental downlink spectrum, promotes interoperability, and enhances competition. The 35x35 MHz band plan that T-Mobile proposed creates more high-value spectrum for competitive bidding and wireless competition with fewer design trade-offs and interference hazards than other alternatives. Most commenters also recommend that the Commission, in crafting forward auction rules, consider the effect that the nation s two largest wireless carriers could have on 1 Expanding the Economic and Innovation Opportunities of Spectrum Through Incentive Auctions, Notice of Proposed Rulemaking, GN Docket No , FCC (rel. Oct. 2, 2012) ( NPRM ). 1

10 forward participation if competitive safeguards are not implemented. Furthermore, a broad consensus has emerged around the need for the reverse auction to provide clear guidelines so that broadcasters have ample incentive to participate and to ensure that bids remain both accurate and binding. Finally, several commenters recommend sound steps to ensure that the repacking is conducted quickly and efficiently to ensure the spectrum resources freed by the incentive auction become available for wireless broadband use as quickly as possible. II. THE PROPOSED 600 MHZ BAND PLAN The 600 MHz band plan represents perhaps the single most critical component to the success or failure of the incentive auction. Selecting the right band plan involves the balancing of innumerable competing goals in which the overly dogged pursuit of any one objective risks compromising or even thwarting the others. Push too aggressively on interference avoidance and the amount of available broadband spectrum could plummet. Stress expansion capacity too much and device costs could soar. Offer too many air interface options and inefficiencies could sap broadband capacity and destroy interoperability. Amidst these band plan design choices, one point of general consensus has emerged: every commenter discussing the issue has recommended that the Commission first address the 84 megahertz of spectrum above Channel 37 and below Channel Focusing first on the 84 2 See, e.g., Comments of CTIA The Wireless Association, GN Docket No , at 21 (Jan. 25, 2013) ( CTIA Comments ); Comments of the National Association of Broadcasters, GN Docket No , at 45 (Jan. 25, 2013) ( NAB Comments ); Comments of Sprint Nextel Corporation, GN Docket No , at 7 (Jan. 25, 2013) ( Sprint Comments ); Comments of Verizon and Verizon Wireless, GN Docket No , at 7 (Jan. 25, 2013) ( Verizon Comments ); Comments of AT&T Inc., GN Docket No , at 32 (Jan. 25, 2013) ( AT&T Comments ); Comments of Sony Electronics, Inc., GN Docket No , at 3 (Jan. 25, 2013) ( Sony Comments ); Comments of Belo Corp., GN Docket No , at 18 (Jan. 25, 2013) ( Belo Comments ); Comments of Comcast Corporation and NBCUniversal Media, LLC, GN Docket No , at 20 (Jan. 25, 2013) ( Comcast & NBCU Comments ); Comments of Google Inc. and Microsoft Corporation, GN Docket No , at 32 (Jan. 25, 2013) ( Google and Microsoft Comments ); Comments of Motorola Mobility, LLC, GN Docket No , at 9 (Jan. 25, 2013) ( Motorola Comments ); Comments of National Cable & Telecommunications Association, GN Docket No. 2

11 megahertz of spectrum above Channel 37 allows the Commission to maximize the number of high-value, five-megahertz paired spectrum blocks available for auction, creates the greatest opportunity for competitive entry and expansion, relies on cost-effective technology, and imposes the fewest number of compromises in design and performance. Allocating any spectrum cleared beyond 84 megahertz for supplemental downlink also helps carriers economically satisfy consumer demand for downlink capacity without creating additional interference or compromising other important Commission objectives. Alternative band plans that do not maximize the number of paired spectrum blocks above Channel 37 offer too few opportunities for competitive entry and expansion, set unrealistically high (or low) expectations for state-of-the-art technology, or impose needlessly restrictive limitations on technical innovation and performance. Focusing first on the spectrum above Channel 37 and specifically maximizing the number of five-megahertz pairs in this band segment would allow the Commission to create a solid foundation of high-value paired spectrum with reasonable levels of design and interference tolerance. A. Widespread Support Exists for a Number of Core Elements in the 600 MHz Band Plan. 1. Maximize Paired Spectrum. The initial round comments reflect broad agreement that maximizing the amount of paired spectrum would increase the utility and value of the 600 MHz spectrum. 3 Maximizing the , at 7 (Jan. 25, 2013) ( NCTA Comments ); Comments of Qualcomm Incorporated, GN Docket No , at 4 (Jan. 25, 2013) ( Qualcomm Comments ); see also Ex Parte Letter from AT&T, Inc., Intel Corporation, National Association of Broadcasters, Qualcomm, T-Mobile, and Verizon Wireless to Gary Epstein and Ruth Milkman, GN Docket No (Jan. 24, 2013) ( Band Plan Principles Joint Letter ). 3 See Comments of T-Mobile USA, Inc. GN Docket No at iii, 10, 13; Comments of Alcatel-Lucent, GN Docket No , at (Jan. 25, 2013) ( Alcatel-Lucent Comments ); AT&T Comments at 2, 18-19; Comments of Cellular South, Inc., GN Docket No , at 6 (Jan. 25, 2013) ( C Spire Comments ); Comments of the Competitive Carriers Association, GN Docket No , at 13 (Jan. 25, 2013) ( CCA Comments ); 3

12 amount of paired 600 MHz spectrum through uplink and downlink bands in paired 5x5 MHz blocks as T-Mobile proposes (with the uplink channel located in the upper channels adjacent to the Lower 700 MHz band) would yield a number of benefits. First, maximizing paired spectrum would accelerate deployment of wireless broadband services to the public. As the initial round comments overwhelmingly show, paired spectrum is not only the industry preference, but also the industry standard, supported by the existing technology infrastructure. 4 Licensing the 600 MHz spectrum in paired 5x5 MHz blocks would allow wireless providers to leverage the existing technology to offer services in the 600 MHz band as soon as possible. 5 As noted by AT&T, most long-term evolution ( LTE ) providers today use Frequency Division Duplexing ( FDD ) technologies and therefore require separate, Comments of the Consumer Electronics Association, GN Docket No , at ii, 4, (Jan. 25, 2013) ( CEA Comments ); CTIA Comments at 20, 22; Comments of Leap Wireless International, Inc. and Cricket Communications, Inc., GN Docket No , at 5 (Jan. 25, 2013) ( Leap Comments ); Comments of MetroPCS Communications, Inc., GN Docket No , at 21 (Jan. 25, 2013) ( MetroPCS Comments ); Qualcomm Comments at i-ii, 2; Comments of Research in Motion Corporation, GN Docket No , at 8 (Jan. 25, 2013) ( RIM Comments ); Verizon Comments at v, See, e.g., MetroPCS Comments at 21 ( Paired blocks generally are the strong preference of the industry, and are critical to support new entrants into a market. ); CEA Comments at 20 (explaining that because [m]ost mobile broadband technologies operate on paired spectrum allocations, with one block dedicated to uplink communications, and the other dedicated to downlink communications, paired allocations would best facilitate the deployment of new wireless broadband services ); RIM Comments at 8 (explaining that, because existing commercial standards are designed for paired spectrum operation, the pairing of spectrum assignments within the 600 MHz band is an important objective for the efficient deployment of services ). 5 Many commenters agree that the Commission should auction paired spectrum in generic categories of five megahertz blocks. See CEA Comments at ii, 4, 18-19; MetroPCS Comments at 19-21; Qualcomm Comments at 5, 20-21; RIM Comments at 6-7; Alcatel-Lucent Comments at 24-25; AT&T Comments at 2; CCA Comments at 12; C Spire Comments at 6; CTIA Comments at 20, 22; Leap Comments at 5; Motorola Comments at 13; T-Mobile Comments at iv, 13-14; Verizon Comments, at v, 6, As Motorola explains, five megahertz blocks will align with a variety of wireless broadband technologies, including Wideband-Code Division Multiple Access (W- CDMA), High Speed Packet Access (HSPA), and perhaps most importantly LTE (when 5 megahertz blocks are aggregated to form 2x10 blocks). Motorola Comments at 13; see also Alcatel-Lucent Comments at 25; CCA Comments at 12; CEA Comments at 18; C Spire Comments at 6; CTIA Comments at 20; MetroPCS Comments at 20; RIM Comments at 6; T-Mobile Comments at 14; Verizon Comments at 15. The Consumer Electronics Association highlights that fungible, 5x5 MHz paired blocks would best enable providers to assemble the amount of spectrum they need to offer mobile broadband. CEA Comments at 4; see also RIM Comments at 6. As AT&T explains, offering generic five-megahertz blocks will greatly simplify the auction process, and it will also ensure denser (and thus more efficient) competition for the spectrum assets at issue. AT&T Comments at 41; see also Verizon Comments at 16; Motorola Comments at 10. 4

13 dedicated uplink and downlink spectrum to provide LTE service. 6 Further, even beyond LTE, paired spectrum will be critical for future wireless broadband technologies. As Alcatel-Lucent explained, paired spectrum will better prepare carriers in implementing and offering next generation technologies for which the broadband demands currently are unknown. 7 Second, maximizing the amount of paired 600 MHz spectrum would enhance competition in the mobile broadband marketplace. 8 As explained by Alcatel-Lucent, the need to acquire paired spectrum will be especially acute for new entrants and other carriers with limited spectrum holdings because it offers these carriers the opportunity to obtain much needed lowfrequency spectrum. 9 Paired 600 MHz spectrum plays an especially important role in promoting competition because it offers extremely good propagation characteristics that can provide competitive operators and new entrants especially those serving non-urban areas with opportunities to efficiently deploy new services. 10 Accordingly, the Commission should use this proceeding as an opportunity to adopt rules and policies that spur competition and lighten the multitude of barriers that wireless providers currently face to deploy service. Third, maximizing paired spectrum best meets the needs of wireless broadband operators and represents a superior alternative to unpaired, time-division duplex ( TDD ) band plans supported by a handful of commenters. Prior to filing comments in this proceeding, Huawei predicted that there will be very little interest in devices that support just LTE TDD within the 6 AT&T Comments at Alcatel-Lucent Comments, at See CCA Comments at 13; C Spire Comments at 6; MetroPCS Comments at 21; T-Mobile Comments at 5. 9 Alcatel-Lucent Comments at C Spire Comments at 7; see also CCA Comments at 13 (explaining that paired spectrum will result in faster, more efficient deployment of 4G services, and therefore promote competition ). 5

14 mobile operator community. 11 And with the sole exception of Sprint and Sprint s subsidiary Clearwire, every wireless broadband operator filing comments in this proceeding sought an FDD allocation in the licensed 600 MHz band spectrum. 12 Given the overwhelming support for an FDD band plan, the Commission can reasonably expect bidders to favor an FDD band plan over TDD alternatives and bid correspondingly more money at auction for FDD than TDD spectrum. Recent economic analysis supports this conclusion. In 2011, for example, the Brattle Group economic consulting firm found that pairing the AWS-3 spectrum with spectrum in the 1755 MHz band would generate approximately $12 billion at auction, compared to just $3.6 billion if the AWS-3 band were auctioned on an unpaired basis. 13 More recent studies from the same authors affirm and expand this conclusion, finding that the present value of profits from deploying unpaired spectrum is expected to be 40% lower than deploying paired spectrum, other things equal. 14 Those findings are also consistent with a recent Huawei white paper, which surveyed auction results and concluded that [m]obile operators have historically been able to purchase TDD spectrum at a lower price than FDD primarily because there are limited options for using TDD spectrum. 15 Although auction revenues are not a primary goal of the 600 MHz proceeding, they offer a potentially helpful indicia of market value. Here, the record demonstrates that the vast preference for FDD will better meet the expectations of wireless broadband providers and, in so doing, generate more revenue than TDD band plan alternatives. 11 Daryl Schoolar, LTE TDD Goes Mainstream: Mobile Ecosystem Puts Support Behind the Standard 19 (Nov. 2012), available at ( Huawei TDD Report ). 12 Sprint s vendor Alcatel-Lucent also indicated that TDD operations might merit consideration. Alcatel-Lucent Comments at The Brattle Group, The Economic Basis of Spectrum Value: Pairing AWS-3 with the 1755 MHz Band is More Valuable than Pairing it with Frequencies from the 1690 MHz Band 1-2 (Apr. 11, 2011), 14 Coleman Bazelon & Giulia McHenry, Spectrum Value 16 (Aug. 28, 2012), available at 15 Huawei TDD Report at 18. 6

15 The overwhelming majority of wireless operators want the maximum possible FDD allocation in the 600 MHz band. The parties also generally agree that unpaired allocations such as a supplemental downlink offering should occur only after the Commission exhausts the possibilities for paired spectrum Maximize Opportunities for Competition During and After the Auction. The record also reflects widespread agreement that the Commission should adopt rules and policies in this proceeding that promote competition during and after the auction. 17 Competition from rivals with sufficient scale motivates larger firms to provide competitively priced services and make investments in quality and innovation. 18 Competition among wireless carriers also encourages competition in complementary downstream markets, such as wireless infrastructure, mobile broadband devices, mobile applications, and the other goods and services that comprise the large and immensely important mobile data sector CTIA Comments at 20 ( Given the desirability of paired spectrum, and based on the analysis to date, the Commission should emphasize pairing spectrum bands and should not allocate spectrum for supplemental downlink unless no pairing option is feasible. ); Alcatel-Lucent Comments at (urging the Commission to adopt a band plan that makes as much paired spectrum available as possible, and noting the myriad reasons why potential bidders would prefer to acquire paired blocks over unpaired blocks); Verizon Comments at 17 (stating that the Commission should license any cleared spectrum that cannot be efficiently licensed in paired spectrum blocks for supplemental downlink ); Qualcomm Comments at 15 ( [I]n areas of the country where additional TV broadcast spectrum is recovered, that spectrum should be used to support SDL operations. ); C Spire Comments at 7 ( [T]he Commission [should] offer unpaired spectrum only after the Commission has paired as many blocks as possible in a market. ). 17 For example, the Competitive Carriers Association has asked the Commission to reaffirm its longstanding commitment to inclusive auction design that includes structural features to ensure broad and balanced participation by a wide range of interested parties, consistent with its statutory directives. CCA Comments at 4. Likewise, C Spire urged the Commission to adopt a band plan that promote[s] competition and the broadest possible deployment of mobile broadband services. See C Spire Comments at C Spire Comments at 2-6; Comments of Free Press, GN Docket No , (Jan. 25, 2013) ( Free Press Comments ); Comments of Public Interest Spectrum Coalition, GN Docket No , (Jan. 25, 2013) ( PISC Comments ); Sprint Comments at 7-10; U.S. Cellular Comments at 30-34; Comments of the Writers Guild of America, West, Inc., GN Docket No (Jan. 25, 2013) ( Writers Guild Comments ); CCA Comments at See Writer s Guild Comments at 2-4; U.S. Cellular Comments at 31-32; Free Press Comments at 14. 7

16 An essential prerequisite for competition is providing enough spectrum to support competitors. 20 Incorporating at least 35x35 MHz of spectrum not only increases the number of the high-value paired spectrum blocks offered during the competitive bidding process, but also provides an opportunity for three operators in every market to acquire the 20 megahertz of paired spectrum (10x10 MHz) to support a high-capacity, high-efficiency mobile broadband service. 21 Achieving at least three operators in every market affords consumers greater choice in wireless carrier and moderates upward pricing pressure by ensuring that at least one other competitor has spectrum space alongside the two largest carriers. 22 In contrast, proposals in which only two licensees have the opportunity to acquire 10x10 MHz, such as proposals to adopt smaller pairings of 25x25 MHz or 20x20 MHz, create the potential for the two dominant carriers to further consolidate their spectrum share and to engage in anti-competitive behaviors, such as tacit signaling and accommodating that can reduce consumer surplus. 3. Conduct a Timely Auction. Numerous commenters joined T-Mobile in supporting the Commission s goal of meeting the nation s pressing demand for additional broadband spectrum by conducting the 600 MHz incentive auction in Given the burgeoning demand for additional wireless broadband spectrum, 24 the Commission should view skeptically suggestions to take as much time as 20 See PISC Comments at 67; C Spire Comments at 6-7; Sprint Comments at 3; CCA Comments at T-Mobile Comments at U.S. Cellular Comments at 33-34; Sprint Comments at See, e.g., CTIA Comments at 8; T-Mobile Comments at 60; U.S. Cellular Comments at 53; CEA Comments at 17; Comments of Mobile Future, GN Docket No , 2 (Jan. 25, 2013) ( Mobile Future Comments ); Verizon Comments at Comments of Cisco Systems, Inc., GN Docket No , 4 (Jan. 25, 2013) (Cisco Comments); CTIA Comments at 8-11; Motorola Comments at 4; PISC Comments at 12; Verizon Comments at 2-3; Comcast & NBCU Comments at 38; RIM Comments at 3; MetroPCS Comments at 1-2; Comments of the Telecommunications Industry Association, GN Docket No , 3 (Jan. 25, 2013) ( TIA Comments ). 8

17 necessary to get the auction done right. 25 As many commenters note, consumers and the economy cannot afford delay. 26 Perhaps more to the point, precisely what constitutes getting the incentive auction right will vary by commenter, and the sheer dint of time will not forge consensus where none exists. No one seriously contemplates undue haste in resolving the complex issues the Commission must address in this proceeding. Reasoned decision-making, however, does not require an inordinate amount of time. Nor does delay itself produce a more sound or just result. 27 The Commission has issued a detailed set of proposals and has received equally detailed comments in response. Although the issues presented in this proceeding are complex, the hundreds of comments submitted in this proceeding, combined with the reply comments and a growing body of ex parte submissions, will aid the Commission s timely resolution of this proceeding. The extensive record evidence should allow the Commission to fulfill its goal of auctioning additional wireless broadband spectrum next year. 4. Establish a Fixed Amount of Downlink Spectrum. Commenters also generally support a 600 MHz band plan that incorporates a fixed amount of downlink spectrum, regardless of the amount of spectrum that is cleared on a marketby-market basis. Licensing downlink spectrum in this manner would reduce handset complexity 25 NAB Comments at CTIA Comments at 4-12; Verizon Comments at 1-2; AT&T Comments at 14; Sprint Comments at 1; Cisco Comments at 4; CEA Comments at 6-12; MetroPCS Comments at 1-2; TIA Comments at 3-4; US Cellular Comments at 2; CCA Comments at See, e.g., CTIA Comments at 8; T-Mobile Comments at 60; U.S. Cellular Comments at 53; CEA Comments at 17; Mobile Future Comments at 2; Verizon Comments at 56; see also Hélène Landemore, Majority Rule and the Wisdom of Crowds: the Task-Specificity of Majority Rule as a Predictive Tool (Aug. 17, 2010), available at (noting that once the creative and brainstorming process of generating new ideas has reached diminishing returns, there is a role for majority rule, which is not merely second best next to deliberation, but its necessary complement ). 9

18 and the need for the wireless standards bodies to adopt multiple band plans for the licensed 600 MHz spectrum. As noted by Motorola, a band plan with defined uplink and downlink frequencies[] should apply across the entire country regardless of whether some markets are not fully cleared, because, under that approach, only one new 3GPP band would need to be profiled. 28 Likewise, AT&T has encouraged the Commission to define sets of contiguous downlinks blocks to enable[e] the industry to use a single passband (and a single duplexer) for any block within such a set. 29 T-Mobile agrees. A consistent amount of nationwide downlink spectrum would lower the costs of deploying a network, promote competition, and reduce the risk of consumers acquiring devices that are incapable of operating throughout the nation. 30 Further, committing to establish a uniform amount of downlink spectrum regardless of how much uplink spectrum is available could alleviate some of the uncertainty surrounding the precise band plan that will be implemented following the reverse auction and, as CEA notes, enable forward auction participants to plan and budget for deployment, handset procurement, and meeting consumer demand Create Supplemental Downlink Spectrum for Asymmetric Pairing. Many commenters joined T-Mobile in supporting the Commission s proposal to license as supplemental downlink any 600 MHz spectrum that cannot be paired. 32 Licensing unpaired spectrum for supplemental downlink would allow carriers to meet the current and future demand 28 Motorola Comments at AT&T Comments at Id. 31 CEA Comments at See, e.g., Verizon Comments at 17; AT&T Comments at 33; CTIA Comments at 21; Qualcomm Comments at 15; RIM Comments at 9; CEA Comments at

19 for downlink capacity efficiently. Mobile subscribers consume much more data than they create. 33 As a result, downlink traffic has increased substantially relative to uplink traffic and this traffic imbalance appears likely to continue for several years. 34 Creating a supplemental downlink responds to these asymmetrical demands on wireless networks while recognizing that dedicated uplink spectrum already exists across many different frequency bands, including the 700 MHz, 800 MHz, 850 MHz, 1.9 GHz, 2 GHz, 2.1 GHz, and 2.3 GHz bands. With supplemental downlink, carriers can combine the 600 MHz band frequencies with existing spectrum resources in other bands to satisfy downlink-intensive data traffic demand without introducing new inefficiencies into their existing spectrum portfolios. 35 A supplemental downlink configuration limits the need for additional uplink spectrum relative to TDD and poses fewer technical complications than TDD, both by eliminating the need for inefficient guard bands and by avoiding the added complexity of coordinating the precise data timing configurations of multiple parties. 36 In short, allocating unpaired spectrum for supplemental downlink would allow carriers to better leverage existing spectrum resources in other bands in order to quickly and efficiently boost aggregate mobile broadband capacity for consumers. 33 Verizon Comments at 17; Qualcomm Comments at 16; Sprint Comments at 19-20; RIM Comments at 9; CEA Comments at 20; Clearwire Comments at Verizon Comments at 17; Qualcomm Comments at 16 (noting that mobile broadband traffic data shows that the ratio of downlink to uplink can be 10:1 or greater ). 35 Verizon Comments at 17; AT&T Comments at 33, Qualcomm Comments at 16-17; CTIA Comments at 21; CEA Comments at See discussion infra at Part II. D

20 B. Commenters Agree that the Commission s Lead Band Plan Proposal Suffers from Shortcomings that Other Band Plans Can Address. The comments highlight two primary problems with the Commission s proposed band plan. 37 First, the breadth of frequencies encompassed by the Commission s lead band plan would require the development and use of at least two low-frequency antennas. 38 While antenna-related concerns will arise anytime the Commission seeks to license a substantial amount of spectrum at auction, they will prove especially acute if the Commission adopts its lead proposal for the 600 MHz band. For most antennas used in mobile wireless communications, antenna size is inversely related to the frequency; that is, the lower the frequency, the larger the antenna needed for efficient transmission. Thus, separating the uplink and downlink bands to the extent proposed by the Commission would eliminate the possibility of a single antenna covering the entire range of frequencies in the 600 MHz band. As Qualcomm noted, in the case of the 600 MHz spectrum, a single antenna may not be able to support a band plan in which downlink spectrum is located below Channel 37 and paired with uplink spectrum directly adjacent to the Lower 700 MHz band. 39 As a result, equipment manufacturers would have to design handsets that include either an external antenna or make additional changes in handset design to accommodate two antennas. According to Qualcomm, [a]dding an additional antenna into smartphones and tablets to support this band increases device size, complexity, and cost, and 37 CTIA Comments at 21, 22, 25; Verizon Comments at 8-9; Motorola Comments at 9; Qualcomm Comments at 6; RIM Comments at 8; AT&T Comments at 5, 20, 25, 30; Alcatel-Lucent Comments at 14-15; Comments of Nokia Siemens Networks US LLC, GN Docket No , 10 (filed Jan. 25, 2013) ( Nokia Siemens Comments ). 38 See, e.g., Motorola Comments at 9; Qualcomm Comments at 6; RIM Comments at 8; Verizon Comments at 8; AT&T Comments at 5, 20, See Qualcomm Comments at 6; see also RIM Comments at 8 (cautioning that the Commission s proposal to offer downlink spectrum at 608 MHz would create a large duplex separation that will have an adverse impact on antenna design); T-Mobile Comments at 9 (relating antenna concerns expressed to T-Mobile by equipment manufacturers). 12

21 introduces a number of material design challenges given consumer demand for wireless devices with smartphone-sized form factors. 40 Second, placing high-power television broadcasters in the duplex gap would produce an elevated risk of intermodulation interference, 41 which occurs when two or more signals combine to produce unwanted interference at new frequencies. 42 In this case, high-power television broadcast stations would be likely to interfere with low-power broadband user equipment, while base stations and cellular broadband operations would interfere with sensitive television receivers. Although the intensity of distortion will vary with the magnitude of the two signals, the more frequencies that are mixed together (and at higher powers), the more interference is generated. 43 Accordingly, and as noted by Verizon, leaving broadcast operations in the duplex gap will increase the risk of harmful interference against which current mobile device and base station filter technology cannot protect. 44 This potential for interference poses a serious challenge to implementation of broadband operations in the 600 MHz band. 45 While these challenges should prove manageable through sound engineering practices, overcoming interference would require considerable technical ingenuity and additional expenditure of time and capital compared to less complex alternatives. Whereas some of the alternative band plans proposed in the record would incorporate TV 40 Qualcomm Comments at See, e.g., Alcatel-Lucent Comments at 14-15; AT&T Comments at 25; CTIA Comments at 25; Nokia Siemens Comments at 10; Verizon Comments at AT&T Comments at CTIA Comments at & n.71 ( Intermodulation products are categorized according to order and can result from the interaction of two or more frequencies. The greater the number of frequencies involved, the greater the number of intermodulation products generated. ). 44 Verizon Comments at 19; see also AT&T Comments at 25 (asserting that placement of multiple television stations in the duplex gap could cause substantial interference in the 600 MHz, 700 MHz, and PCS receive bands and substantially degrade mobile operations in those bands ) (emphasis in original). 45 See, e.g., AT&T Comments at

22 stations in the duplex gap under some clearing scenarios, the Commission s proposal would incorporate TV stations in the duplex gap in nearly all clearing scenarios. 46 Alternative approaches that reduce the number of scenarios in which broadcasters will operate in the 600 MHz duplex gap offer less costly and complex designs that can achieve better performance. T-Mobile agrees with the many commenters recommending that the Commission first pursue alternative band plans that focus on pairing some or all of the 84 megahertz of spectrum above Channel 37 and below Channel So long as the Commission can establish at least 35 megahertz of paired spectrum using an alternative band plan, it should. C. The 35x35 MHz Band Plan T-Mobile Has Proposed Harnesses the Benefits of the Commission s Lead Band Plan While Avoiding Many of the Risks. Establishing a 35 megahertz uplink allocation at MHz paired with a 35 megahertz downlink allocation at MHz maximizes both the total amount of paired broadband spectrum available for competitive bidding in the 600 MHz band and the efficiency of the wireless broadband allocation created. 48 Of course, every spectrum band plan requires a series of trade-offs among spectrum efficiency, antenna efficiency, interference risk, cost requirements, and handset design allowances, among other considerations. And the 35x35 MHz configuration that T-Mobile has proposed for the 600 MHz band plan is no exception. In the case of a 35x35 MHz band plan, however, the issues are few and readily managed through 46 NPRM 126, See, e.g., Nokia Siemens Comments at 11 (noting that the Commission can combat the risk of intermodulation interference caused by TV operations in the duplex gap by adopting a band plan that would avoid any TV transmissions in the FDD duplex gap and relocat[ing] as many remaining TV channels as possible to the lower part of the spectrum below Channel 37 and fill in all of the band above Channel 37 with mobile broadband ); see also CTIA Comments at 21; NAB Comments at 45; Sprint Comments at 7; Verizon Comments at 7; AT&T Comments at 32; Sony Comments at 3; Belo Corp. Comments at 18; Comcast & NBCU Comments at 20; Google and Microsoft Comments at 32; Motorola Comments at 9; NCTA Comments at 7; Qualcomm Comments at 4; Band Plan Principles Joint Letter. 48 See T-Mobile Comments at

23 existing technical solutions. Moreover, reasonably anticipated technical advances make achieving a 35x35 MHz configuration even more feasible in 2014 or 2015 than it already is today. 49 Taken together, a 35x35 MHz band plan offers the largest amount of paired wireless broadband spectrum with the fewest and most readily managed band-specific design requirements. The 35x35 MHz band plan T-Mobile has proposed features 35 megahertz of downlink spectrum paired with 35 megahertz of uplink spectrum separated by a ten megahertz duplex gap. 50 In markets where less than 84 megahertz (or 14 TV channels) are cleared, this plan would Figure 1: T-Mobile's proposed 35x35 MHz band plan maximizes paired spectrum while minimizing guard bands. prioritize downlink spectrum over uplink spectrum and preserve a common duplex gap in the 600 MHz band. In markets where more than 84 megahertz is cleared, T-Mobile s band plan 49 Most of the band plans advanced in this proceeding envision supplemental downlink in some portion of the spectrum below the paired frequencies above Channel 37. The addition of supplemental downlink to the 600 MHz band will require antennas and filters to extend for roughly 110 MHz below 698 MHz. The active tuning, surface mounted and printed circuit board implementations necessary to support supplemental downlink in an economical manner will assist in the much more manageable challenge of developing a single set of antennas and filters to support the 35x35 MHz band plan. 50 See T-Mobile Comments at

24 would make the additional spectrum available for supplemental downlink in five megahertz channels located below Channel 37. Similar to every other variation of a Down from Channel 51 band plan proposed in this proceeding, the 35x35 MHz band plan would: Provide for a common amount of downlink spectrum to minimize device complexity despite variable amounts of cleared broadcast spectrum; Eliminate the need for inefficient guard band spectrum between the 600 MHz and 700 MHz bands; Prioritize downlink spectrum over uplink spectrum to accommodate growing downlink traffic; and Avoid the risk of harmful intermodulation interference associated with placing highpower television stations in the duplex gap as contemplated under the Commission s lead band plan. T-Mobile s 35x35 MHz proposal also offers several distinct features from other Down from Channel 51 band plans. The most notable of these features is the capacity of a 35x35 MHz band plan to offer more paired wireless broadband spectrum, more efficiently than any other comparable band plan. As explained in greater detail in the attached technical analysis prepared by Roberson and Associates, LLC ( Roberson Technical Analysis ), a 35x35 MHz configuration optimizes the spectrum for the most widely used, most readily available form of 4G broadband in existence: FDD LTE technology. 51 The Roberson Technical Analysis calculates that by eliminating inefficient gaps and segments a 35x35 MHz configuration increases nominal wireless broadband capacity by 40 percent compared to a 25x25 MHz band plan. 52 Moreover, the active bandwidth of a 35x35 MHz band plan is larger than the active bandwidth of a 25x25 MHz band plan. 51 Roberson and Associates, LLC, Analysis of the 35x35 MHz Band Plan Proposal for 600 MHz Spectrum (Mar. 11, 2013) ( Roberson Technical Analysis ), attached as Exhibit A. 52 Id. at 3,

25 Assuming both bands use the same size duplex gap to separate uplink and downlink spectrum, the duplex gap in the 35x35 MHz band comprises a smaller portion of total active bandwidth under a 35x35 MHz band plan than under a 25x25 MHz band plan. As a result, the spectrum usage efficiency of the 35x35 MHz band plan stands at 88% compared to only 83% with a 25x25 MHz band plan. 53 Segmenting the bandwidth above Channel 37 into separate downlink-only allocations or increasing the available duplex gap would introduce inefficiencies and capacity constraints. By removing spectrum from the most valuable paired uses, band plans of less than 35x35 MHz not only decrease opportunities for competitive entry, but also reduce the total amount of valuable paired spectrum resources available for competitive bidding. Few other band plans aside from the 35x35 MHz configuration advanced by T-Mobile establish up to seven paired, fivemegahertz spectrum blocks available for auction, or allow up to three paired, ten-megahertz spectrum blocks available for auction. These additional paired blocks and the increased potential to aggregate multiple paired blocks into larger blocks of more efficient contiguous spectrum offer the wireless industry a wealth of spectrum ideally suited for LTE broadband operations that promises to encourage broad-based participation in the 600 MHz auction, promote intense activity among competing bidders, and increase competition in the wireless broadband marketplace following the close of the auction. Technical challenges raised about a 35x35 MHz band plan are readily addressed. The three primary challenges of implementing a 35x35 MHz band plan are: (1) antenna performance; 53 Id. at

26 (2) radiofrequency duplex filter feasibility; and (3) harmonic interference. 54 Each of these issues is manageable through the use of sound engineering and the deployment of current and near-term technologies. 1. Antenna Performance. Antenna design involves fairly complex considerations of size, volume, material, scale, and tuning range to produce a product with optimum electromagnetic resonance to cover the proper combination of frequencies while meeting smartphone size constraints. And yet two truisms of antenna design remain: first, using one antenna is less costly than using two antennas, and, second, using a smaller antenna is less costly than using a larger one. To avoid the cost and complexity of multiple antennas or larger antennas, a few commenters offer 20x20 MHz or 25x25 MHz band plans above Channel While these band plans will avoid the use of multiple, larger antennas, these approaches yield considerably less spectrum for wireless broadband use than the 35x35 MHz band plan that T-Mobile supports. Based on the findings in the Roberson Technical Analysis, moreover, the Commission does not need to reduce the total amount of paired wireless broadband spectrum available for competitive bidding to achieve these efficiencies. 56 Instead, carriers can use the same antenna that supports a 25x25 MHz configuration to support a 35x35 MHz configuration with little or no performance degradation and few if any additional costs To be sure, other technical and design challenges exist, but these are not unique to the 35x35 MHz band plan that T-Mobile has proposed. Design challenges common to all Down from Channel 51 band plans are addressed in detail in the Roberson Technical Analysis. See id. at AT&T Comments at 32; Qualcomm Comments at Roberson Technical Analysis at Id. at

27 Antenna length depends on the center frequency of operation and the total range of operation: the larger the range of operation, the bigger the antenna. With a total operating spectrum range of 80 megahertz, the passband of the 35x35 MHz band plan is 20 megahertz larger than the 60 megahertz operating spectrum range of the 25x25 MHz band. This larger operating spectrum range means that the antenna needed to support a 35x35 MHz configuration will be larger than the antenna needed to support a 25x25 MHz configuration. But the Roberson Technical Analysis shows that the difference is manageable through use of current and emerging technologies. Assuming a worst-case traditional passive antenna implementation, the length of the antenna used in a 35x35 MHz configuration would only be about seven millimeters longer than the length of the antenna used in a 25x25 MHz configuration. 58 This much additional space roughly the equivalent of just seven grains of sugar can be accommodated within existing smartphone designs. 59 But even if the additional volume were somehow too large for end-user devices, carriers could still use the smaller antenna that is optimized to support the 25x25 MHz configuration and experience only modest radiofrequency efficiency losses of db. 60 Moreover, carriers could likely compensate for all of these modest losses by simply increasing transmission power. 61 Perhaps best of all, existing, off-the-shelf technology, such as active antennas and printed antennas, can handle the requirements of a 35x35 MHz band plan very 58 Id. 59 Id. at Designers typically scale down internal antennas to meet the space requirements of a smartphone; therefore, real-world deployments would create a de minimis difference in antenna size. See Roberson Technical Analysis at 16 & n Id. at Id. LTE incorporates power control, which allows user equipment to adjust dynamically as it approaches the base station. As a result, the only user equipment that transmits at the maximum power of 23 dbm will occur near the cell edge or in other, low-coverage conditions. Hence, carriers can increase the transmit power for a significant number of users without compromising the system. Id. at

28 effectively without any performance losses all while occupying only about half the volume of the worst-case traditional passive antennas. 62 Thus, the Commission need not lower its spectrum auction target below 35x35 MHz to achieve the antenna efficiencies associated with a 25x25 MHz band plan Duplex Filter Performance. Duplex filters are a critical component in the path between the antenna and the transmitter or receiver amplifier. Duplex filters for wireless broadband operations today consist of one of two basic technologies: Surface Acoustic Wave ( SAW ) or Film Bulk Acoustic Resonator ( FBAR ). With either SAW or FBAR technology, state-of-the-art performance allows these filters to span approximately four percent of the center frequency of operations, which in the case of the 600 MHz band translates to approximately 28 megahertz. 64 This span is expected to expand to at least 30 megahertz by 2014 or The problem with a 35x35 MHz configuration is that existing duplex filters are five megahertz too small to cover the full uplink or downlink bands. Yet this problem has a straightforward solution: deploy two overlapping filters, each of which would cover two-thirds of the overall bandwidth. 65 While this dual filter configuration would require additional switches, operators in other markets have demonstrated that the technique is both realistic and 62 Id. at Id. at A 25x25 MHz band plan with supplemental downlink in the remaining spectrum will experience exactly the same modest and easily resolved antenna performance constraints as a 35x35 MHz band plan. Carriers using either band plan would presumably want one antenna rather than two. In designing an antenna, however, the relevant factor is not the air interface or directionality, but rather the span of spectrum covered and the particular frequencies involved. Therefore, AT&T s suggestion that a 25x25 MHz or 20x20 MHz band plan with supplemental downlink has different antenna performance demands than a 35x35 MHz band plan would appear to be incorrect. See AT&T Comments at 6. Unless AT&T intends to introduce another antenna for the supplemental downlink, its 25x25 MHz band plan plus supplemental downlink would have precisely the same antenna constraints as a 35x35 MHz band plan without supplemental downlink. Roberson Technical Analysis at Id. at Id. at

29 cost effective. Qualcomm, for example, has identified commercially feasible solutions in the Asia Pacific Telecommunity (APT), which allow for a 45x45 MHz band plan in the MHz band ten megahertz more than contemplated by T-Mobile in the 600 MHz band. 66 Figure 2: An excerpt from Qualcomm's 2011 Harmonization of the Digital Dividend demonstrates how overlapping filters allow for the realization of 45 megahertz pass bands. Dual duplexer configurations, such as the plan APT adopted in 2010, 67 are fairly costeffective because aside from the additional duplexer and associated switch, the rest of the hardware, including the amplifier and antenna, can be shared. 68 As a result, the two filters and switch would not consume substantially more space or impose substantially more costs than a single duplex filter would. Specifically, an additional filter and switch would consume only about 12 mm 3 more volume in 600 MHz device than a single filter while the cost of the additional hardware for two filters as opposed to one filter, especially at scale volumes, would be insignificant. 69 Finally, the Roberson Technical Analysis indicates that at the current steady rate 66 See Qualcomm, Harmonization of the Digital Dividend, (May 2011), 67 See Asia-Pacific Telecommunity, Report on Harmonised Frequency Arrangements for the Band MHz, No. APT/AWF/REP-14 (Sept. 2010), see also 4G Americas, The Benefits of Digital Dividend at 16 (September 2012) ( 4G Americas Report ), (explaining that [c]hannel bandwidths up to 15 MHz can be supported anywhere within the [Asia-Pacific] band, but channel bandwidths of 20 MHz are limited to the upper and lower parts of the band and may not be employed in the mid-portion of the band where the filters overlap ). 68 Roberson Technical Analysis at Id. 21

30 of technical progress in SAW and FBAR filter technology, the Commission can reasonably anticipate that a 35 megahertz filter will prove commercially available in As 4G Americas noted, [a]dvances will be made in technology and components that will drive down cost and improve the chances for increased economies of scale across multiple bands and radio formats. 71 Thus, the migration path to a single, smaller, lower-cost filter appears imminent and, in any case, should occur well prior to the auction and subsequent repacking and clearing of the 600 MHz band. If necessary in the meantime, the dual-filter approach would not only deliver more paired wireless broadband spectrum for competitive bidding in the 600 MHz auction, but also promote more competitive wireless broadband offerings. The use of two overlapping filters, especially when combined with the random or quasi-random assignment of spectrum blocks discussed below, would ensure that no operator is left outside of the scale economies that might otherwise apply to only one of two non-overlapping filter ecosystems. The dual-filter configuration would also promote interoperability throughout the 600 MHz band, which helps ensure no consumer has to buy a new iphone or other costly device simply to transfer from a carrier with frequencies that use one 600 MHz duplex filter to another carrier with frequencies that use another 600 MHz duplex filter. In short, the interim use of two duplex filters to achieve a 35x35 MHz band plan does not impose material costs on carriers, nor does the use of dual filters require inordinately complex design solutions. On the contrary, the use of dual filters during the interim period until a single 35 megahertz filter is developed would deliver substantial benefits by assuring that all carriers 70 Id. 71 4G Americas Report at

31 have an opportunity to compete amidst a constantly changing global ecosystem in device technology that places considerable value on volume purchases. Together with the additional and more efficient configuration of wireless broadband spectrum available for competitive bidding that the 35x35 MHz band plan achieves, these benefits outweigh any modest, short-term incremental costs associated with using two duplex filters. 3. Harmonic Interference. The final consideration associated with a 35x35 MHz band plan above Channel 37 is the potential for harmonic interference. Some commenters have claimed that allowing any mobile uplink transmissions between MHz would generate harmful third-order harmonic interference into portions of the PCS band ( MHz) and harmful fourth harmonic interference into portions the BRS/EBS band ( MHz). 72 These arguments are unpersuasive. T-Mobile has no dispute with these commenters calculations that operations in certain portions of the 600 MHz would generate third- and fourth-order harmonic effects in portions of other spectrum bands used for broadband services. Third- and fourth-order harmonic effects always exist at three and four times the fundamental frequency of any transmission See, e.g., Nokia Siemens Comments at 14; CTIA Comments at 26; AT&T Comments at 27; Qualcomm Comments at 7-8. As explained by Qualcomm, harmonic interference from a transmit signal is caused by nonlinear characteristics inherent to the output stages of transmitters and occurs at multiples of the transmitted signal (e.g., a 695 MHz signal has a second order harmonic at 1390 MHz, a third-order harmonic at 2085 MHz and so on). See Qualcomm Comments at 8. Consequently, a harmonic signal could fall within the passband of a receiver within the same or nearby device, and potentially degrade the receiver s performance. Id. 73 Other types of harmonic interference will occur fourth, fifth, sixth, seventh, and so on but, like waves emanating from a stone thrown into a pond, these conditions grow progressively weaker as they become more and more removed from the original source and, therefore, pose much less of a concern. Roberson Technical Analysis at Most of these additional harmonics do not, in any case, fall within bands used for wireless broadband communications. Id. at

32 The signal that produces the third harmonic will generally not be that strong, however. With a maximum power output of 23 dbm, a 600 MHz device transmitter in the user device will inject -26 dbm in a PCS receiver in a nearby device, which has the potential to result in up to seven decibels of desensitization to the PCS receiver. 74 But this condition represents the worst case. The vast majority of the time, the device transmitter will operate with far less than 23 dbm power and, as a result, produce far less desensitization into the PCS receiver. 75 Moreover, the internal guard bands that 600 MHz licensees will use to protect against adjacent channel interference will further limit actual power into the device. 76 Finally, the fourth order harmonic is more attenuated and even less likely to interfere with communications signals than the third harmonic is. 77 While devices that incorporate both the interfering and victim frequencies could experience harmful interference under high-power conditions, operators have several techniques to prevent harmonic interference from ever occurring. 78 These simple, cost-effective interference-avoidance techniques include improved filtering, careful block selection, and spectrum exchanges. 79 As Sprint noted in its comments, [p]otential third-harmonic conflicts 74 Id. at Id. 76 Id. 77 While the fourth harmonic falls within a portion of the 2.5 GHz BRS-EBS band, the effect of this harmonic is reduced an additional 10 db as compared to the third harmonic. The result is a signal of -103 dbm, which is well below the interference limit guideline of -100 dbm, and should not affect receivers in the BRS-EBS band. Id. at Encapsulating the interfering and victim signals in the same device is itself a worst-case scenario. Whatever the harmonic effects within the same device, those effects are even less substantial between different devices. The additional separation distance and attenuation that occurs between different devices, among other things, further reduces the potential for harmful interference. See Nokia Siemens Comments at See id. 24

33 already exist in the U.S., and yet we have seen little evidence of such interference problems to date. 80 In short, harmonic effects do not pose a serious interference concern. Sprint s lack of concern about harmonic effects is especially noteworthy. Sprint holds a nationwide license in the MHz band and through its Clearwire subsidiary controls most of the 2.5 GHz band in many geographic locations. These Figure 3: An excerpt from the Roberson Technical Analysis shows the location of third harmonic effects in the PCS bands. bands represent precisely where the strongest harmonic interference effects from the 600 MHz band would fall. Thus, the beneficiary of AT&T s ostensible concern is not principally AT&T, but rather Sprint, which, quite reasonably, remains unconcerned. While the risk of harmonic interference is remote and easily avoided, the effect of capping paired spectrum at fifty megahertz as AT&T has proposed is not: excluding the 600 MHz frequencies capable of producing harmonic interference in small portions of the PCS and BRS bands will limit the pool of high-value paired spectrum available at auction. Thus, while AT&T correctly notes that forgoing an auction of the 600 MHz uplink frequencies that produce the harmonic would eliminate any possibility of interference, 81 AT&T s solution takes a sledgehammer to a nail. Solving the possibility of harmonic interference by not auctioning the potentially offending spectrum would impose considerable costs on consumers and taxpayers in 80 Sprint Comments at See AT&T Comments at 32, 34 (offering a band plan proposal that would not create uplink spectrum below Channel 47). 25

34 response to a fairly limited, readily managed concern. Moving spectrum from high-value paired configurations to lower-value unpaired (or, worse, guard-band) configurations represents a lastresort solution suitable for consideration only after all other avenues that could allow the most productive use of the available spectrum resource have been exhausted. The national imperative of maximizing the amount of mobile broadband spectrum available through the competitive bidding process simply should not, in Sprint s words, be stymied by technical concerns that can be solved through equipment design and operating practice. 82 Where, as here, the potential for harmonic interference is limited and manageable and the benefits of maximizing paired spectrum are extensive and broad-based, the Commission can and should take note of successful, real-world spectrum-management practices and auction the maximum amount of paired for commercial wireless broadband use. D. Band Plans that Provide Less than a Potential of 70 Megahertz of Paired Spectrum for Auction Will Decrease Auction Revenues and Harm Competition. Absent compelling evidence of pervasive harmful interference, little justification exists for the Commission to adopt a band plan that offers less than 70 megahertz of paired spectrum. Paired spectrum is simply too valuable not to maximize the amount available, and if the Commission does not adopt a band plan capable of supporting as much paired spectrum as possible, competition during and after the 600 MHz auction will likely suffer. 1. AT&T and Qualcomm s Band Plan. AT&T and Qualcomm advance broadly similar band plans that offer only 50 megahertz of paired spectrum. As shown in Figure 4 below, the 25x25 MHz configuration would prohibit 82 Sprint Comments at

35 uplink operations below Channel 47 and establish a separate downlink-only band with its own antenna below Channel 41: 83 Figure 4: AT&T's proposed band plan would cap paired spectrum at 50 megahertz. Although AT&T indicates its band plan is notional and subject to change, AT&T s longtime technical consultants, Jeffrey H. Reed and Nishith D. Tripathi, advance a narrow set of options that never seem to result in more than fifty megahertz of paired spectrum available for auction in the 600 MHz band. 84 The Commission should reject the AT&T and Qualcomm band plan for at least three reasons. First, a 35x35 MHz band plan offers 40 percent more nominal capacity than a 25x25 MHz band plan as well as more options for highly efficient aggregations of contiguous 83 AT&T Comments at 32; Qualcomm Comments at 15. The duplex gap shown in Figure 4 is 14 megahertz in width, but smaller configurations could be employed. This larger gap accommodates AT&T s proposal to use the duplex gap for supplemental downlink operations. AT&T Comments at See AT&T Comments, Exhibit A, Jeffrey H. Reed and Nishith Tripathi, The 600 MHz Spectrum Auction: An Analysis of the Band Plan Framework 1-5, 25-37; accord Qualcomm Comments at 3 n.5 (discussing, but ultimately rejecting, a 30x30 MHz or 35x35 MHz plan as unwise ). 27

36 bandwidth. While the base units of LTE operations are five megahertz spectrum pairs, using ten megahertz pairs would deliver material gains in efficiency, throughput and performance. 85 With demand for wireless broadband speed and efficiency increasing annually, carriers will strive to acquire spectrum in ten megahertz pairs when possible. 86 The 50-megahertz cap on paired spectrum available at auction envisioned by AT&T and Qualcomm is, therefore, significant because it would effectively limit the number of ten megahertz pairs in any market to just two bidders a pair of winners that in all likelihood would include Verizon and AT&T, but exclude T-Mobile, Sprint and the rest of the competitive carriers. Because contiguous spectrum blocks seem likely to play such a large role in current and future deployments, adopting the AT&T/Qualcomm proposal risks further consolidation of the already highly concentrated wireless market. Second, maximizing the amount of high-value paired spectrum available for competitive bidding will generate funds for important public projects such as the broadcast relocation, the FirstNet public safety broadband network, and deficit reduction. Like many other commenters, AT&T and Qualcomm acknowledge that carriers value paired spectrum and its efficiencies more highly than the unpaired spectrum that they incorporate into their band plans. 87 Maximizing the amount of paired spectrum simply recognizes this reality and configures the 600 MHz band in a manner designed to capture the most intense interest from the vast majority of likely bidders. 85 See NPRM 130 n.208; RIM Comments at 7 ( [W]ider blocks handle traffic more efficiently than the minimum bandwidths, and new spectrum should be designed to meet growing demand in the most efficient way possible. ). 86 As Research in Motion recognizes, bandwidth in excess of a 5 MHz uplink and a 5 MHz downlink will be essential, in many circumstances, to provide the expected levels of service in today s increasingly mobile society. RIM Comments at While AT&T s concept of incorporating additional supplemental downlink channels inside the duplex gap is intriguing, the concept is not unique to a 25x25 MHz configuration and, if useful, could apply equally as well to any FDD configuration, including a 35x35 MHz configuration that would offer more high-value paired spectrum for competitive bidding. 28

37 Third, the ostensible technical concerns that AT&T and Qualcomm associate with a 35x35 MHz band plan do not pose any significant obstacles today and will pose even fewer obstacles tomorrow. One of the hallmarks of the wireless industry is its tremendous capacity for innovation. 88 AT&T and Qualcomm, however, both take an exceptionally dim view of the capacity of available and emerging technologies, the multiplicity of design choices available from global equipment vendors, and the ingenuity of network engineers to design and operate high-capacity, high-performance, cost-effective communications systems in the face of complex radiofrequency environments. Indeed, Qualcomm goes so far as to say that the Commission should base the 600 MHz band plan only upon off-the-shelf device components available on the market today, rather than components available next year, when the spectrum will be auctioned. 89 According to Qualcomm, taking into account ongoing technical progress in the dynamic and innovative wireless market would impair the value of the spectrum by creating unspecified implementation risks for wireless operators. 90 This position is a novel one for Qualcomm, which typically emphasizes the rapid innovation found in the wireless industry. 91 According to Qualcomm s Chairman of the Board and Chief Executive Officer, Paul Jacobs, for example, Qualcomm s vision through the years has been clear to look past the horizon and, in doing so, anticipate and drive future capabilities 88 See, e.g., Letter from Steve Largent, President and CEO, CTIA the Wireless Association, to Hon. Lamar Smith, Chairman, House Judiciary Committee (Nov. 28, 2012), available at STEM%20Jobs%20Act.pdf ( The hallmark of the wireless industry is innovation. ); CTIA, Innovation and Competition (last visited Feb. 18, 2013), available at (noting that the U.S. wireless industry is the most innovative and competitive in the world. ). 89 Qualcomm Comments at Id. 91 See, e.g., Qualcomm, Inc., 10-Q (filed Jan. 30, 2013), ( Our industry is subject to rapid technological change, and we must make substantial investments in new products, services and technologies to compete successfully. ). 29

38 and uses of wireless technologies. 92 The Commission should put its faith in Qualcomm s business leadership. Technological change is ever-present and rapidly expanding the capacity of wireless companies to accommodate new spectrum allocations. than capable of achieving a 35x35 MHz band configuration. Existing technology is more Even if it were not, the Commission should devise a band plan based on both existing and reasonably anticipated technological developments. 2. Google and Microsoft. Although they do not offer formal band plans, several commenters, including Google and Microsoft, suggest that the Commission should maximize the amount of unlicensed spectrum in the band plan. 93 These parties suggest that the Commission could create a massive duplex gap of up to 28 megahertz in addition to large guard bands for use by unlicensed devices. 94 As an initial matter, like the AT&T/Qualcomm band plan, maximizing unlicensed spectrum would significantly harm competition in the wireless marketplace. For example, under the Google/Microsoft proposal, the maximum paired spectrum available above Channel 37 would be 25x25 MHz. 95 As explained in the preceding section, any band plan that offers less 92 Dr. Paul E. Jacobs, Chairman of the Board and Chief Executive Officer, Qualcomm, Empowering Communities Worldwide Through 3G (2010) available at brochure-english.pdf 93 See Google and Microsoft Comments at 32; Comcast & NBCU Comments at 44; Free Press Comments at 13; PISC Comments at Google and Microsoft Comments at 32, Because there is only 84 megahertz of spectrum above Channel 37, a 2x25 MHz band plan with a 28 MHz duplex gap is only possible if a six megahertz guard band is employed above Channel 37; a larger guard band would further reduce the availability of paired spectrum. 30

39 than 35x35 MHz of paired spectrum risks further consolidation of the most valuable sub-1 GHz spectrum by the two largest carriers. 96 The Google-Microsoft and similar proposals also directly contravene the Spectrum Act s directive that guard bands shall be no larger than is technically reasonable to prevent harmful interference between licensed services outside the guard bands. 97 In adopting this provision, and, indeed, through the very creation of the incentive auction framework, Congress expressed a desire to maximize spectrum available for commercial wireless. 98 If the Commission adopted proposals to maximize unlicensed spectrum, it would directly contravene Congressional intent. Proponents of maximizing unlicensed use wrongly seek to stretch the language of the Spectrum Act to accommodate their proposals. For example, Comcast claims that by allowing the Commission to adopt technically reasonable guard bands, Congress employed statutory language that permits the Commission to consider other policy goals including facilitating 96 To the extent commenters seek an unlicensed allocation from spectrum acquired in the incentive auction, valuable frequencies below 1 GHz are not the appropriate location. The Commission has recognized that spectrum below 1 GHz is an especially scarce and an especially valuable resource due to its superior propagation characteristics. See, e.g., Policies Regarding Mobile Spectrum Holdings, Notice of Proposed Rulemaking, WT Docket No , FCC (rel. Sept. 28, 2012); see also Scott Wallsten, Is There Really a Spectrum Crisis? Quantifying the Factors Affecting Spectrum License Value, Technology Policy Institute at 20 (Jan. 23, 2013) (indicating that, based on analysis of every spectrum auction since 1996, spectrum below 1 GHz, when used for broadband, is more valuable than spectrum above 1 GHz ). In circumstances where spectrum scarcity exists, such as under 1 GHz, the Commission s Spectrum Policy Task Force explained that the exclusive use model is most effective at balancing [competing] claims to a scarce resource because it is best able to create significant incentives for efficient spectrum use. FCC Spectrum Policy Task Force, Report of the Spectrum Efficiency Working Group at 33 (Nov. 15, 2002), Unlicensed devices, which typically operate at low power and operate over short distances, are better suited to spectrum in the 5 GHz range due to the reduced risk of interference and the relative abundance of spectrum available. 97 Middle Class Tax Relief and Job Creation Act of 2012, Pub. L (b) (Feb. 22, 2012) ( Spectrum Act ). 98 Attempts to sideline valuable spectrum for unlicensed use can take various forms. Proponents of unlicensed operations not only propose excessively large guard bands, but also recommended large white spaces between television channels. See Google and Microsoft Comments at 49. If TDD were adopted, it seems reasonable to presume unlicensed advocates would want unreasonably large guard times as well as unreasonably large guard bands. All of these proposals should be rejected as directly contravening both the plain language and the intent of the Spectrum Act. 31

40 unlicensed use. 99 That is not so. Congress did not just require guard bands to be technically reasonable, but instead required guard bands to be no larger than is technically reasonable to prevent harmful interference between licensed services outside the guard bands. 100 Comcast s interpretation also directly conflicts with the explicit legislative history, which makes clear that [e]nlarging the guard bands for any reason other than mitigating interference, such as facilitating unlicensed use, would conflict with section 6407(b). 101 As the Commission itself explains, the use of guard band is intended to minimize interference between dissimilar adjacent operations. 102 Under the Spectrum Act, unlicensed uses should not take precedence over commercial licensed uses. For their part, Google and Microsoft suggest that guard bands and duplex gaps are technically reasonable if there is any potential for interference. 103 This interpretation impermissibly reads the plain language no larger than technically reasonable out of the statute. 104 Further, the statute s limit would be meaningless under this interpretation: base stations transmit low level noise and potentially interfering signals across a very wide range of frequencies. Google and Microsoft s any interference interpretation would effectively allow for guard bands nearly as large as the amount of spectrum freed at auction. As commenters recognize, moreover, the Spectrum Act significantly constrains the Commission s 99 Comcast & NBCU Comments at Spectrum Act 6407(b). 101 Majority Committee Staff, Committee on Energy and Commerce, U.S. House of Representatives, Memorandum: Hearing on Keeping the New Broadband Spectrum Law on Track, at 4 (Dec. 10, 2012) (emphasis added). 102 See NPRM Google and Microsoft Comments at See Astoria Federal Savings & Loan Ass n v. Solimino, 501 U.S. 104, 112 (1991). (explaining that statutes should be interpreted so as to avoid rendering superfluous any statutory language). 32

41 decisionmaking authority as to permissible sizes for duplex gap and guard bands. 105 For example, as TIA explains, reasonable means that guard bands cannot be excessive or extreme in size. 106 The legislative history supports this interpretation. The version of the bill that originally passed the House did not even mention guard bands; the technically reasonable restriction was later added to restrict the Commission s discretion. 107 Moreover, the record demonstrates that a smaller duplex gap than the 28 megahertz gap proposed by Google and Microsoft is technically reasonable. 108 As the Commission notes, Band 8 ( MHz and MHz) has a ten megahertz duplex gap. 109 Other duplex gaps that illustrate what would be no larger than technically reasonable include the 11 megahertz gap for Band 20 ( MHz and MHz bands), and the 13 megahertz gap for Band 12 ( MHz and MHz band (lower 700 MHz)). 110 When enacting the Spectrum Act, Congress was acting with these technically reasonable duplex gaps (as well as other technically reasonable guard bands) in mind. 111 Google and Microsoft do not contend that any of these duplex gaps are not technically reasonable (because they are). Nor do they directly explain how 105 Cisco Comments at 12; MetroPCS Comments at 24; TIA Comments at TIA Comments at 9 n TIA Comments at 9-10 n.32. Free Press, however, citing to a blog article, contends that Congress reportedly considered and rejected... language mandating that the guard bands be no larger than technically necessary. Free Press Comments at 5 n.8 (citing Stephan E. Coran, Congress Makes Sweeping Changes to Spectrum Policy; Authorizes TV Band Incentive Auctions, Telecom-Media-Tech Law Blog (Feb. 21, 2012), Neither Free Press, nor the blog article, nor any other commenters, cite to where this discussion occurred in the legislative history. In any event, the language technically necessary was never part of any official version of the bill. 108 Qualcomm Comments at 15; Sony Comments at 4; Verizon Comments at 18; AT&T Comments at 32; CTIA Comments at NPRM NPRM 167 n NCTA asserts that the duplex gap is not subject to the technically reasonable limit, arguing that the duplex gap is not a guard band. NCTA Comments at 11. The duplex gap, however, is a specific type of guard band under the Spectrum Act it is a guard band that protects against harmful interference between licensed uplink and downlink outside the duplex gap. See Spectrum Act 6407(b). Indeed, other commenters, including Google and Microsoft, explicitly recognize that the limits of section 6407(b) apply to the duplex gap. Google and Microsoft Comments at 38; Comcast & NBCU Comments at

42 a duplex gap larger than these technically reasonable gaps can be no larger than technically reasonable. Google and Microsoft nevertheless suggest that if the Commission s goal is to auction generic, fungible blocks, the Commission must use larger guard bands and a larger duplex gap. 112 But the Commission need not and, as a practical matter, cannot create technically identical blocks of spectrum. However desirable the goal, perfectly fungible spectrum is impossible, and its absence will not affect auction revenue or bidder behavior in ways that matter to the outcome of the auction. 113 Bidders will always have idiosyncratic preferences that make some spectrum bands more valuable to some bidders than others. As Alcatel-Lucent notes, the Commission would find it impossible to account for all such individualized situations without completely abandoning interchangeability and dispensing with competitive bidding as a license distribution mechanism. 114 Moreover, not all differences in spectrum blocks are wholly idiosyncratic. Each block of spectrum will operate at a distinct frequency and will face a different composite noise floor level that will result in unique performance characteristics. So long as the resulting performance does not suffer from harmful interference, variations among blocks are both inevitable and wholly inconsequential. As CTIA notes in the context of broadcast channels, 112 Google and Microsoft Comments at 39 (suggesting that the absence of interference is required for the Commission to implement a band plan with spectrum blocks that are as similar and technically interchangeable as possible ) (quoting NPRM 152). 113 Although Google and Microsoft themselves are unconcerned with fungibility of spectrum and the presumable increase to auction revenues they are merely advocating for maximal unlicensed allocation AT&T contends that concerns over fungibility and interference loom large in ultimate auction revenues. See AT&T Comments at As AT&T argues, [i]f bidders face uncertainty about the value of the spectrum they will ultimately receive... they will discount their bids, thereby decreasing auction revenues. AT&T Comments at 41, 42; see also Qualcomm Comments at 20 ( [F]or the auction to be successful, each 2 x 5 MHz FDD block should be spectrally identical to each other, which requires the FCC to incorporate sufficient guard bands so that the spectrum blocks adjacent to the guard bands are protected the same as the non-adjacent spectrum blocks. ). 114 Alcatel-Lucent Comments at 19 n.19. Alcatel-Lucent notes that secondary market where blocks are traded in aftermarket transactions can readily account for idiosyncratic values for different spectrum blocks. Id. 34

43 [c]hanges in UHF frequency will necessarily mean that there will be changes in coverage that are unavoidable. 115 To illustrate the concept, CTIA offers the example of different coverage areas achieved using different frequencies in the related context of replicating broadcast area coverage: A higher UHF channel may lead to less coverage, while a lower UHF channel could increase coverage. 116 Because replicating a broadcaster s signal following the repack will prove virtually impossible, CTIA concludes that the Commission should allow ample leeway for other frequencies to satisfy the statutory directive to replicate the performance of the station s original channel. 117 A similar level of tolerance should apply to spectrum blocks available in the forward auction. It simply is not possible for the Commission to achieve perfect substitutability among blocks. No two bidders will value the identical block of spectrum the same, and no two blocks of spectrum will possess precisely the same coverage or signal carrying capacity. Fungibility, in other words, is necessarily a matter of degree. 118 A quest for perfect fungibility would likely cause the auction to produce less paired spectrum than simply allowing for some variations among blocks. With limited spectrum available, achieving true fungibility would require the Commission to allocate more spectrum for guard bands, which, in turn, would likely force a reduction in the amount of paired spectrum available for broadband use a result far more detrimental to competition, consumers, and the broadcast incentive auction than having blocks that vary somewhat in technical characteristics. 115 CTIA Comments at Id. 117 Id. 118 See, e.g,. Martin Weiss, et al., When is Electromagnetc Spectrum Fungible?, IEEE Dynamic Spectrum Access Networks (Aug. 2012), available at (evaluating numerous factors affecting the substitutability of spectrum and concluding that a multidimensional analysis of substitutability is required because we can only speak of two bands being fungible along a continuum ). 35

44 Most wireless broadband providers, including T-Mobile, would much rather have more spectrum available for competitive bidding than have fewer spectrum blocks available at auction with fewer differentiations among them. Tolerance for variations among blocks will produce more spectrum for mobile broadband deployment and more high-value blocks subject to competitive bidding. The Commission should ignore the fungibility red herring, reject the sprawling guard bands and duplex gaps contemplated by Google-Microsoft, and adopt only those guard bands that are no larger than technically reasonable to prevent harmful interference as the Spectrum Act requires. 3. A TDD Band Plan Has Serious Shortcomings, and Commenters Overwhelmingly Support an FDD Band Plan. The vast majority of commenters prefer paired spectrum suitable for next generation FDD deployment, and for good reason. 119 FDD represents the industry standard for nextgeneration networks, and allocating spectrum for TDD use would create unnecessary complexities that would lead to inefficient use of the 600 MHz band. 120 Although TDD may merit consideration elsewhere, 121 the significant drawbacks of any TDD plan ultimately demonstrate it to be a far inferior option to paired downlink spectrum in the 600 MHz band. 119 See, e.g., AT&T Comments at 18-19; C Spire Wireless Comments at 6; CTIA Comments at 21; Google and Microsoft Comments at 32; MetroPCS Comments at 21; Motorola Comments at 9; NAB Comments at 45; NCTA Comments at 7-10; Nokia Siemens Comments at 11-12; Qualcomm Comments at 4; RIM Comments at 8; Sony Comments at 4; Verizon Comments at Verizon Comments at 17-18; AT&T Comments at Alcatel-Lucent at 3; see also Sprint Comments at 22; CCA Comments at (recognizing that most carriers will employ FDD but noting that the Commission should not foreclose consideration of proposals for TDD use in the 600 MHz band ). 36

45 To be sure, the TDD air interface, which divides its uplink and downlink traffic by time rather than frequency, offers some benefits. 122 The contiguous nature of the TDD allocation, for example, means there is no need for a duplex gap a noteworthy efficiency. 123 Moreover, it would be easier to incorporate TDD rather than FDD in some 600 MHz clearing scenarios because the variable amounts of spectrum that the 600 MHz reverse auction will clear will not always neatly support both ends of a paired configuration. 124 Finally, TDD can be customized to support asynchronous traffic as opposed to having fixed amounts of uplink and downlink established through the band plan design. 125 Unfortunately, the costs of adopting a TDD configuration in the 600 MHz band greatly outweigh these benefits. First, conducting TDD operations in the presence of FDD operations would require additional guard bands within the 600 MHz band, which would limit the amount of spectrum that could be auctioned for mobile broadband applications. 126 As Alcatel-Lucent notes, TDD would require separations of at least ten megahertz at every intersection of TDD with FDD uplinks, FDD downlinks, or broadcasting operations. 127 Second, even if the entire 600 MHz band were dedicated to TDD an outcome few but Sprint and its subsidiary Clearwire support the plan would still require an extra guard band between 600 MHz TDD operations and the Lower 700 MHz that FDD configurations do not require Sprint Comments 17-21; Clearwire Comments Sprint Comments at 4; Clearwire Comments at Clearwire Comments at Sprint Comments at 19-20; Clearwire Comments at See Verizon Comments at 17; Nokia Siemens Comments at Alcatel-Lucent Comments at Sprint Comments at 22; cf. Verizon Comments at 8; T-Mobile Comments at

46 Third, avoiding interference among TDD operators would require all TDD operators in any given area to carefully synchronize their operations. 129 TDD operators, which can customize the ratio of uplink to downlink transmissions, 130 would have to agree to a common uplinkdownlink ratio and then agree to carefully calibrate and continuously monitor their operations to ensure that the alternation of base station transmissions and end user transmissions occur at precisely the same times. While Clearwire has successfully managed to coordinate the timing of TDD operations in the high-frequency 2.5 GHz band that it largely controls, it remains unclear whether or to what extent competing 600 MHz operators operating could overcome the barriers to negotiation and accomplish the same feat. 131 Fourth, the very propagation characteristics of the 600 MHz band that make the spectrum so attractive to FDD operators namely, the band s ability to penetrate deep inside buildings and travel long distances may make the 600 MHz band less desirable for TDD use. Research in Motion explains that the long range (wide area) coverage of the 600 MHz band is most suited to FDD operation due to the excessive propagation delays for channel turnaround encountered in TDD systems at these ranges. 132 Whereas FDD configurations rely on fixed guard bands, TDD configurations must rely on guard times to separate uplink transmissions from downlink transmissions enough to avoid harmful interference. These guard times are generally equal to the time it takes for a signal to travel to and from the base station plus a round trip delay TDD uses primarily ten megahertz blocks and, unless carefully synchronized, each operational block would require guard bands on either side of their operations to separate operational blocks. 130 A TDD air interface divides the data stream into frames and, within each frame, assigns different time slots to the uplink and downlink transmissions. A TDD air interface allows carriers to customize the percentage of uplink and downlink they support. 131 Sprint Comments at 20 n Comments of Research in Motion, Docket No , at 8 (Jan. 24, 2013). 133 Moonblink, TDD vs. FDD and WiMax (last visited March 4, 2013), 38

47 Thus, even in the absence of guard bands, the requisite guard times of TDD will incorporate inefficiencies into 600 MHz band TDD operations that will constrain their carrying capacity and throughput. Fifth, unlike the guard bands associated with FDD configurations, the guard times associated with TDD configurations are ephemeral and, at the present time, much less practical for use by unlicensed devices. The guard times, in other words, introduce idle spectrum capacity that is far less capable of being put into use by the large number of companies that have sought unlicensed access to the guard bands associated with FDD configurations. For these reasons, among others, the Commission should not pursue a TDD allocation in the 600 MHz band. III. THE FORWARD AUCTION A. The Record Supports the Adoption of a Spectrum Cap that Would Limit the Amount of Spectrum Below 1 GHz that a Single Licensee Can Hold. The record reflects widespread agreement that the Commission should adopt rules in this proceeding to curtail the further consolidation of spectrum suitable for mobile broadband applications. 134 A bright-line rule governing spectrum concentration would (1) promote competition in services that use or rely on wireless spectrum; (2) reduce the administrative costs and delays associated with a case-by-case basis review; (3) prevent efficiency-sapping distortions in bidding that result from not knowing whether or how divestitures will occur; (4) encourage auction participation; and (5) potentially increase auction revenues See C Spire Comments at 3; Free Press Comments at 14-15; PISC Comments at 63, 68; Sprint Comments at 3, 9; U.S. Cellular at CCA Comments at 9; C Spire Comments at 2-6; Jonathan B. Baker, Establishing Auction Rules that Promote Competition in Wireless Services, at 3-4 (Mar. 12, 2013) ( Baker Economic Analysis ), attached at Exhibit B. 39

48 Wireless carriers are not alone in urging the Commission to institute a mechanism that will ensure diversity among holders of valuable spectrum resources, promote long-term competition, and increase auction participation; rather, support for these measures extends to distant downstream competitors, such as the Writers Guild, that recognize the threat that steadily increasing concentration in the wireless industry poses to their businesses. 136 Likewise, other countries have identified spectrum concentration as a threat to the timely and cost effective deployment of broadband services. Last week, for example, Canada s conservative government led by Prime Minister Stephen Harper, who The Economist has recognized as a longstanding advocate of encouraging entrepreneurship and keeping the government s influence on people s lives to a minimum, 137 found that access to spectrum represents a critical barrier to entry in this industry and held that without rules preventing excessive concentration of spectrum holdings, competition could suffer. 138 In adopting spectrum caps for its planned 700 MHz 4G auction, Industry Canada said that the caps would support the objectives of sustained competition and robust investment in a minimally intrusive manner. 139 So too here, the best mechanism available to promote competition is a spectrum cap specifically a cap that would limit any 136 See, e.g., Writers Guild Comments at 4, 5 (noting that it is imperative to address the growing problem of spectrum aggregation because [t]he lack of competition in wireless and the harm it is causing the mobile video market makes it critical that the Commission address spectrum aggregation before making more spectrum available to wireless providers ); Baker Economic Analysis at 5 (noting that excessive spectrum aggregation can encourage the dominant incumbents to frustrate the development of new technologies and business models brought to the market by fringe rivals and potential competitors, including future rivals that cannot now be identified ). 137 Madelaine Drohan, The United States of Canada: the Country Will Look a Lot More Like the Real America, The Economist (Nov. 17, 2011), available at Indeed, U.S. Republican Newt Gingrich has praised Harper as a conservative and pro-american. Michael Bolen, Newt Gingrich: Stephen Harper Gets Shout-Out from Republican Hopeful, Huffington Post (Jan. 24, 2012), See Industry Canada, Policy and Technical Framework: Mobile Broadband Services (MBS) 700 MHz Band, Broadband Radio Service (BRS) 2500 MHz Band, (March 2012), available at Id. 40

49 single licensee from acquiring more than one-third of the available commercial mobile spectrum below 1 GHz in the incentive auction. 1. Excessive Concentration in the U.S. Wireless Market Harms the Public Interest. Although consumer demand for wireless communications services is rising at an unprecedented rate, the amount of spectrum ideally suited for wireless broadband applications below 1 GHz has become increasingly concentrated. The consolidation has been most pronounced among the nation s top two largest carriers, Verizon Wireless and AT&T. As Sprint noted, those two carriers have aggregated approximately 75% of the commercial spectrum below 1 GHz, including 86% of it in the top 10 U.S. markets and over 80% in the top 50 markets. 140 The Commission s own reports on the state of the U.S. wireless industry likewise bear out this marketplace development: in 2011, the Commission reported that Verizon Wireless and AT&T held 67.20% of 700 MHz commercial spectrum, and 91.30% of cellular (850 MHz) spectrum. 141 C Spire likewise noted that, as of 2010, consolidation in the U.S. wireless industry measured nearly 350 points above the threshold of a highly concentrated market, according to the Herfindahl-Hirschman Index ( HHI ). 142 The concentration of low-band spectrum has undermined the public interest in a number of respects. Carriers (such as T-Mobile) that have been unable to acquire sub-1 GHz spectrum have been forced to incur substantially higher costs to deploy an infrastructure that achieves the coverage that consumers demand. Accordingly, as Sprint notes, the eligibility rules adopted by 140 Sprint Comments at See Implementation of Section 6002(b) of the Omnibus Budget Reconciliation Act of 1993, WT Docket No , Fifteenth Report, FCC & Table 27 (rel. June 27, 2011) ( Fifteenth Report ). 142 C Spire Comments at 3. 41

50 the Commission in this proceeding will have a lasting effect on wireless competition, as well as the innovation and economic growth the Commission envisions the incentive auctions stimulating. 143 The concentration of spectrum also has fueled the development of unforeseeable technical hurdles that have impeded service deployments in other spectrum bands. For example, as noted by U.S. Cellular, the lack of interoperability in the Lower 700 MHz band would not have arisen if there had been a greater diversity of license winners in the A, B, and C Blocks from the outset. 144 T-Mobile agrees that had measures been in place to ensure that a diversity of licensees succeeded in acquiring the 700 MHz spectrum, [i]nteroperability would have been a practical necessity... because all of those carriers would have worked together to develop technology and drive a robust ecosystem A Spectrum Cap Below 1 GHz Should Be Applied to Prevent Further Harm Caused by the Consolidation of Valuable Spectrum into the Hands of a Few Players. The risk of further consolidation is great, and that risk is one recognized even by AT&T, which concedes that some measures may be necessary to prevent undue spectrum aggregation. Although AT&T opposes ex ante limits on the spectrum that particular carriers can obtain through this auction, it has acknowledged that some licensees may acquire an amount of spectrum that would bring its total holdings in a market to a level that is determined to threaten competition. 146 As noted by C Spire, unless the Commission institutes structures that prevent the 600 MHz incentive auction process from resulting in further consolidation... the auction 143 Sprint Comments at U.S. Cellular at Id. 146 AT&T Comments at

51 could result in cementing the Bell duopoly, eliminating further competition from the marketplace, and leaving the Commission with no choice but to engage in heavy-handed, public utility style regulation of a wireless industry dominated by just two operators. 147 Eligibility rules in this proceeding prohibiting a licensee from acquiring more than onethird of the available commercial mobile spectrum below 1 GHz, applied on a market-by-market basis, would mitigate the risk of further concentration of beachfront spectrum among only a few very large providers. 148 As explained in the Baker Economic Analysis, spectrum concentration can allow the largest carriers to obtain or maintain downstream market power by keeping spectrum from their rivals. 149 In an auction without caps, the two largest incumbents would have both the ability and incentive to limit competition because the foreclosure value the two largest incumbents gain from keeping competitors out of the band comes in addition to the value of the spectrum for providing wireless broadband service. 150 Targeting spectrum below 1 GHz is especially important because low-frequency spectrum is less commonly available and more highly valued than higher frequency spectrum. 151 Indeed, the Commission has recognized that below 1 GHz spectrum is uniquely valuable for mobile broadband applications. 152 Licensees with higher frequency spectrum must construct more cell sites in a given geographic area to match the signal coverage of a licensee deploying 147 C Spire Comments at Other parties agree. C Spire has asked the Commission to adopt a spectrum screen that accounts for the aggregation of low band spectrum by the largest operators. C Spire Comments at 5. Sprint likewise endorses license eligibility rules that would preclude a licensee from exceeding the proposed low-band spectrum cap of onethird of available spectrum below 1 GHz. Sprint Comments at Baker Economic Analysis at Id. 151 See, e.g. Scott Wallsten, Technology Policy Institute, Is There Really a Spectrum Crisis? Quantifying the Factors Affecting Spectrum License Value at 20 (Jan. 23, 2013) (indicating that, based on analysis of every spectrum auction since 1996, spectrum below 1 GHz, when used for broadband, is more valuable than spectrum above 1 GHz ). 152 Policies Regarding Mobile Spectrum Holdings, WT Docket No , Notice of Proposed Rulemaking, 27 FCC Rcd 11710, (Sept. 28, 2012) ( Spectrum Holdings NPRM ) (citing sources). 43

52 service using 700 MHz, 850 MHz, and now 600 MHz, band spectrum. 153 As a result, a wireless provider may disadvantage rivals (raising their production costs) by denying them access to lowfrequency spectrum, even if high-frequency spectrum can physically substitute for lowfrequency spectrum to some extent with additional capital investment. 154 Moreover, if wireless services can be provided more cheaply using a mix of spectrum from different bands, firms can still disadvantage competitors by monopolizing a specific band even if different bands have different merits and spectrum in one band can to some extent substitute for spectrum in another band with additional capital. 155 The proposed 1 GHz cap is therefore designed to ensure that the spectrum made available in this auction will be distributed in a manner that fosters competition among both large and small wireless providers, which will produce large consumer benefits. The spectrum cap T-Mobile has proposed would not only give effect to the Commission s statutory obligation to avoid excessive concentration of licensees and distribute licenses to a wide variety of applicants, 156 but also would increase participation in the forward auction. T-Mobile agrees with Verizon that maximizing bidder participation should be one of the central goals of the forward auction. 157 But widespread participation will not occur unless the Commission adopts a cap or screen designed to prevent the risk of excessive spectrum concentration in the hands of the one or two largest carriers. Absent clear eligibility rules that limit the amount of spectrum below 1 GHz that a licensee can hold, smaller bidders may simply 153 The need to construct additional cell sites in higher-frequency spectrum to provide the same level of coverage also requires greater upfront capital investment and generates additional reliance on other bottleneck inputs, such as backhaul from the cell site. In some areas, moreover, constructing additional cell sites may prove difficult or impossible due to physical, regulatory, environmental, and other limitations. 154 Baker Economic Analysis at See id. at U.S. Cellular Comments at 32; 47 U.S.C. 309(j)(3)(B). 157 Verizon Comments at

53 assume that defeat is inevitable and choose not to participate in the forward auction. 158 Indeed, by encouraging widespread participation in the auction, the Baker Economic Analysis concludes that spectrum caps have the potential to increase auction revenues. 159 Although the Commission could try to curtail further spectrum concentration by other means, an ex ante rule would be most effective. Specifically, a bright-line spectrum cap would: Provide the certainty necessary for interested parties to plan their participation in the forward auction, as well as their services, technologies, and financing needs; Avoid the administrative burdens of assessing spectrum holdings on a case-bycase basis after the auction, which requires the Commission to expend substantial resources to assess whether a licensee s holdings are contrary to the public interest and may reduce auction revenues as bidders must discount their bids by the possibility of divestiture; Eliminate the costs and inefficiencies that attend the process of forcing licensees to divest certain licenses; Reduce the incentive of carriers to engage in insincere bidding to acquire spectrum simply to prevent other carriers from securing the necessary spectrum resources to effectively compete; and More effectively prevent spectrum concentration than the Commission s ex post divestiture approach, which, as U.S. Cellular notes, has not imposed any meaningful limitations on the amount or type of spectrum that a single party may acquire. 160 In light of the limited amount of spectrum available for mobile broadband applications, as well as the existing concentration of spectrum below 1 GHz in two dominant carriers, the Commission should adopt a pro-competitive cap on spectrum holdings below 1 GHz. Not taking 158 See Baker Economic Analysis at 10 ( Given the non-trivial fixed costs of auction participation, a firm expecting to be outbid could readily be deterred from participating in the auction in the first place. ). 159 Baker Economic Analysis at 19 (emphasis added). 160 U.S. Cellular Comments at 30; Baker Economic Analysis at 11-14; see also C Spire Comments at 3-4; PISC Comments at 66-68; Sprint Comments at 3,

54 into account the incentives to protect and extend dominant market positions would result in an inefficient allocation of the 600 MHz spectrum. 3. The Commission Should Reject the Objections by Verizon and AT&T to the Spectrum Cap. It is no surprise that the Verizon and AT&T the two carriers with the largest sub-1 GHz spectrum holdings oppose a spectrum cap. However, their objections against the spectrumrelated eligibility proposals lack merit. First, spectrum below 1 GHz is either distinctive, as T-Mobile, Sprint, U.S. Cellular, Leap, CCA, and many others contend, or it is not. 161 AT&T, however, claims both that a cap on excessive spectrum below 1 GHz will harm it, but that a lack of access to below 1 GHz spectrum will not harm competitive carriers. According to AT&T, competitive carriers can simply acquire higher frequency spectrum and then spend more on deploying additional infrastructure in lieu of acquiring low-frequency spectrum. 162 The end result, AT&T contends, will be the same: more costly spectrum plus less costly infrastructure will equate to less costly spectrum plus more costly infrastructure. 163 Of course, ample evidence exists to prove this contention false. 164 But even if spectrum above and below 1 GHz were freely substitutable, the cap should have no effect whatsoever on the dominant holders of below 1 GHz spectrum because they could simply acquire high frequency spectrum at a substantial discount and construct more infrastructure, as AT&T claims others can do. The truth is that spectrum below 1 GHz offers unique value and is 161 T-Mobile Comments at 25; Sprint Comments at 2-3; Leap Comments at 7-8; CCA Comments at 2; U.S. Cellular Comments at Comments of AT&T Inc., Docket No (Nov. 28, 2012), Attachment A, Mark A. Israel, & Michael L. Katz, Economic Analysis of Public Policy Regarding Mobile Spectrum Holdings Id. 164 See supra Section III. A

55 not substitutable for spectrum above 1 GHz. But if despite all the evidence to the contrary spectrum below 1 GHz were substitutable for spectrum above 1 GHz, then adopting a below 1 GHz cap would do no harm. AT&T simply cannot have it both ways. Second, contrary to the assertions of Verizon and AT&T, 165 the Commission has increasingly recognized the growing consolidation of spectrum for commercial mobile services, and has even initiated a separate proceeding to examine the adequacy of its current spectrum holdings policies, as they are generally applied in transactions and at auction. 166 The concentration of the highest-value spectrum below 1 GHz is especially pronounced and the 600 MHz auction threatens to reinforce the concentration that the Commission has identified. Rather than squarely address the issue, the two largest wireless operators have simply ignored the Commission s justifiable sensitivity to the increasing consolidation of spectrum ideal for wireless broadband, as well as the statutory mandate of both the Spectrum Act and Section 309(j) of the Communications Act that the Commission foster competition through spectrum diversity. 167 Third, Verizon has hypothesized that eligibility restrictions would result in some of the 600 MHz spectrum being licensed to providers that do not value it the most or who are incapable of deploying service in a timely manner. 168 But that would not be the case. Rather, a spectrum cap would encourage interested parties who might otherwise not have participated in the auction 165 AT&T Comments at 79-80; Verizon Comments at See generally Spectrum Holdings NPRM. 167 In the 1980s, the Commission assigned the first cellular licenses to the predecessors of Verizon and AT&T among others at no cost. With only two licenses per market, competition languished and so too did wireless deployment. Not until the 1990s, when the Commission awarded multiple new spectrum licenses during an auction conducted under a strict spectrum cap, did the United States market see the proliferation of new competitors, including T-Mobile, Sprint, and other wireless carriers that challenged the dominant incumbents with innovative new offerings. 168 See Verizon Comments at

56 to bid for 600 MHz spectrum. 169 There is no reason to think that these additional bidders would have less of an incentive or ability to build-out the spectrum simply because they are smaller or greenfield operators. The general support among regional wireless carriers for a spectrumbased eligibility rule offers compelling evidence of how a spectrum cap would incentivize participation in the forward auction. Moreover, the Commission can adequately protect against licensees that are unable to timely deploy service by adopting and enforcing construction deadlines and performance requirements, as it has done for other licensed spectrum. 170 Fourth, the claim that a spectrum cap would suppress demand for the 600 MHz band and risk not meeting the closing conditions for the auction lacks any basis. 171 As T-Mobile explained in its comments and as reiterated in the Baker Economic Analysis, a spectrum cap can increase participation in a spectrum auction and yield auction proceeds that exceed what would otherwise result without a cap. 172 Opponents of the spectrum cap have marginalized the essential role that widespread participation would play in enhancing the market clearing prices during the forward auction. Moreover, their objection fails to account for the long-term benefits that promoting spectrum diversity would achieve, and disregards the fact that revenue maximization is just one of several goals that the Commission must pursue in this proceeding. In fact, the Spectrum Act expressly directs the Commission to adopt and enforce rules concerning spectrum aggregation that promote competition, but makes no mention of any obligation to maximize revenue See T-Mobile Comments at See NPRM See Verizon Comments at 41-42; see also Ex Parte Letter from Ari Meltzer, Counsel for the Expanding Opportunities for Broadcasters Coalition, to Marlene Dortch, Secretary, Federal Communications Commission, GN Docket No (Jan. 28, 2013). 172 See T-Mobile Comments at 33; Baker Economic Analysis at 3, 4, 7, See id. at 34 (citing Spectrum Act 6404). 48

57 Finally, a cap on spectrum below 1 GHz would not inject uncertainty into the forward auction, as Verizon has conjectured. 174 Rather, as noted above, clear and early spectrum-clearing targets would provide substantial guidance to auction participants as they determine the markets in which they can aggressively compete for 600 MHz band spectrum. For example, licensees with large spectrum holdings below 1 GHz in a particular market would know at the outset the limitations of their ability to acquire substantially more spectrum in that market. Additionally, the overwhelming majority of interested parties would not be affected at all by the adoption of a spectrum cap. Because so much of the spectrum below 1 GHz is held by Verizon and AT&T, the universe of auction participants that would need to modify their bidding strategies in response to a specific cap is limited. After-the-fact divestitures of the type AT&T has proposed would do more harm than good. 175 Not only are mandatory ex post divestitures uncertain, unpredictable, and timeconsuming, they would materially distort bidding in the forward auction as interested parties would have to guess whether, when, and how the largest carriers would reduce their spectrum holdings to less dominant levels. 176 As the Baker Economic Analysis explains, the cost associated with after-the-fact review are potentially substantial because rivals that might have won in the initial auction but were outbid by the large firm (or decided not to bid because they expected to be outbid) may have found work-arounds by the time the winning firm is required to divest. 177 Allowing the two largest carriers to choose both the spectrum that they would divest 174 See Verizon Comments at See AT&T Comments at See Baker Economic Analysis at Id. at 20. After-the-fact review does not necessarily help Verizon and AT&T, either. A dominant incumbent that falsely believes that its acquisition may be rejected, or fails to win because it reduces its bid due to the risk of postauction divestitures will also result in a less efficient spectrum assignment and potentially lower auction revenues. Id. 49

58 and the buyers eligible to purchase the spectrum would exacerbate the problem by providing a mechanism for the party divesting the spectrum to place the resource with the company least likely to pose a competitive threat. 178 Meanwhile, empowering AT&T and Verizon to shuffle assets between them would do nothing to improve competition and, in fact, would likely only reinforce the kinds of collaboration and tacit signaling that can diminish competition. B. Random or Quasi-Random License Assignments Will Clear More Spectrum, Increase Efficiency, Promote Interoperability, and Raise More Revenue in the Forward Auction. Employing the random or quasi-random assignment of 600 MHz licenses within a common license category of paired or unpaired blocks will encourage scale efficiencies, promote interoperability, eliminate the negative consequences some associate with transparent bidding, and perhaps most important increase the likelihood that the incentive auction will close with the maximum possible amount of spectrum dedicated to mobile broadband use. The alternative to random or quasi-random assignment of 600 MHz license blocks is to conduct an additional assignment auction following the incentive auction. 179 A follow-on assignment auction, however, would divert auction revenue from broadcast clearing, complicate and delay license assignment, and risk creating new impediments to consumer choice. A random or quasi-random assignment process found strong support in the initial comment round. 180 AT&T, for example, called for the Commission to avoid reliance on such supplemental bidding and to ensure that the generic-bidding round is the main event Id. 179 While the Commission could also introduce an assignment auction into the forward auction, conducting an assignment auction as part of the forward auction each round could cause delay. 180 See, e.g., AT&T Comments at 42, Verizon Comments at AT&T Comments at

59 Verizon similarly called for assignment of licenses through rules and policies rather than a follow-on auction 182 Verizon even recognized that a degree of randomness is unavoidable in any assignment procedure, and discouraged the Commission from seek[ing] to resolve inconsistent frequency assignments... through competitive bidding. 183 Other commenters agreed. 184 The mechanics of a quasi-random assignment process are straightforward and can be designed using transparent rules to ensure fairness and accountability. Under one approach, the Commission would generate a list of the block winners within each MEA and then randomly assign blocks to each winning bidder in a randomly generated order. 185 To capture the synergies of geographic and spectrum contiguity, the Commission would constrain the random assignment of licenses slightly by adopting transparent rules to ensure that (1) winning bidders with licenses throughout an MEA obtain the same block within that MEA and (2) winning bidders with more than one block within an MEA obtain contiguous spectrum within that MEA. 186 In a notable limitation, these two constraints on randomness would apply only within an MEA, not across multiple MEAs. This limitation represents an important safeguard to prevent a single licensee from obtaining the same frequency blocks everywhere in the country a condition that creates opportunities to increase rivals equipment costs, thwart roaming obligations, and diminish consumers ability to move among different carriers. 182 Verizon Comments at Id. at Motorola Comments at An alternative to Commission assignment would be to allow the winning bidders to choose their blocks according to a randomized priority set by the Commission. Under this arrangement, the Commission would offer winning bidders the choice of specific licenses according to this random list order, again subject to certain constraints designed to allow for some measure of geographic and spectrum contiguity. 186 Alcatel-Lucent, for example, proposes an assignment rule that would guarantee that bidders that win more than one 5 MHz block obtain blocks that are adjacent to each other, permitting the carrier to operate using wider channels and greater throughput than a single 5 MHz block would allow. Alcatel-Lucent Comments at 25; see also Verizon Comments at 46; RIM Comments at 7. 51

60 To illustrate how the rules would function, suppose in a given MEA there are seven generic license blocks sold among three winners. Two of the winners have three licenses each and one has one license. Assuming that the license blocks are numbered A to G contiguously and the bidder with only one license received priority, then the Commission would assign the bidder that won one license to block A or D or G. If G were randomly assigned to the one-block winner, then one of the three-block winners would be either A-C or D-F within that MEA while the other three-block winner would receive the unassigned contiguous blocks. Because a different quasi-random assignment would occur in each MEA, no one winner would be likely to acquire all of the spectrum on a single set of frequencies, yet every multiple license winner could still capture the synergies associated with acquiring contiguous spectrum and common blocks in any given market area. 187 A random or quasi-random assignment procedure achieves several benefits, including: (1) creating durable incentives for market-driven interoperability; (2) resolving concerns about anonymous bidding concerns; (3) establishing a more readily met closing requirement that will maximize the spectrum acquired for mobile broadband use; and (4) raising more revenue in the forward auction. 1. Creating Durable Incentives for Interoperability. Most commenters agree that adopting an interoperability requirement for the 600 MHz band represents the simplest and most effective means of preventing anti-competitive band 187 T-Mobile Comments at At the same time, a random or quasi-random assignment need not present any hurdles for carriers desiring to aggregate multiple blocks of spectrum. As the Commission recognizes in the Notice, it anticipates that it will assign contiguous blocks to bidders that bid for multiple blocks in the same geographic area. NPRM 64. If significant economies exist to having similar frequencies in every MEA, auction winners could use secondary market exchanges to achieve single-block operations, but that eventuality would offer competitors the opportunity to ensure that incompatibilities do not emerge or hamper competition. 52

61 fragmentation at 600 MHz. 188 Requiring interoperability will increase competition among wireless carriers and allow consumers to switch providers without requiring them to purchase costly new mobile phones, tablets, and computers. 189 Adopting random assignment procedures reinforces and extends the interoperability requirement that T-Mobile and many other commenters support by providing durable, market-based incentives for manufacturers to create interoperable devices capable of tuning across all paired spectrum following the 600 MHz auction. 190 So long as a carrier s 600 MHz blocks are distributed randomly through the band, a carrier has no incentive or ability to develop a custom, boutique band class centered around one block of frequencies. 191 As a result, handset vendors will create phones compatible with all providers, which will increase the ability of users to switch providers without switching phones something that is not necessarily in the interest of handset manufacturers, who would just as soon see the consumer buy another costly device, or in the interest of dominant wireless operators, who use increased switching costs to reduce churn off of their networks. These market-driven incentives for achieving interoperability among multiple wireless operators resulting from diversely assigned frequency holding would have many advantages and 188 C Spire Comments at 9 ( [T]he Commission has the power to protect the 600 MHz band from balkanization by requiring interoperability. ); CCA Comments at 16 ( CCA strongly urges the Commission to require interoperability throughout the 600 MHz band, to avoid the problems that have plagued the Lower 700 MHz band. ); Leap Comments at 7 ( [T]he Commission should ensure interoperability across the entire 600 MHz band, to prevent the splintering that has occurred in the 700 MHz band. ); MetroPCS Comments at 28 ( The Commission must ensure that [the problem of interoperability in the Lower 700 MHz band] does not happen again in the 600 MHz band by mandating a single band class across the entire band. ); Comments of the National Telecommunications Cooperative Association, GN Docket No (Jan. 25, 2013) ( Nat l Telecom. Coop. Ass n Comments ) (supporting any measures that would ensure interoperability ); U.S. Cellular Comments at 26 ( [A]s it has done in the past, the Commission must strive to adopt spectrum policies, including an interoperability requirement, that will benefit consumers in unserved and underserved areas. ); T-Mobile Comments at 21 ( The Commission should require interoperability across all paired 600 MHz band channels. ). 189 See, e.g., CCA Comments at 16; U.S. Cellular Comments at 25-26; T-Mobile Comments at NPRM 162 (explaining that one of the Commission s goals in deciding how best to license this wireless spectrum is encouraging interoperability ). 191 See U.S. Cellular Comments at

62 finds strong support in the record. 192 Interoperability decreases the costs of devices because carriers and manufacturers are able to achieve economies of scale. It decreases the expense consumers face when switching carriers and promotes roaming between networks. And it promotes the deployment of mobile broadband services. 193 As the Commission well knows, lack of interoperability is devastating to the value and usefulness of a set of frequencies the Lower 700 MHz A Block, for example, has languished following a decision by one of the largest carriers to consolidate holdings elsewhere in the band and to develop hardware incapable of operating on the 700 MHz A Block frequencies. It is thus unsurprising that many commenters desire interoperability throughout the 600 MHz band. 194 The Commission should adopt an interoperability mandate for the 600 MHz band and ensure a diversely held set of frequency assignments in the band. Together, these measures would help guard against the damaging consequences of artificially limiting common equipment to frequencies associated with only one wireless carrier. 2. Resolving Concerns about Anonymous Bidding. Random or quasi-random assignments can also address smaller and mid-size carriers concerns with anonymous bidding. 195 Several smaller carriers contend that anonymous bidding will allow larger carriers to shut them out of certain blocks, thereby compromising their ability to obtain roaming agreements and interoperable equipment. Unless competitive carriers know 192 Several commenters support creating incentives for interoperability through auction design. While it opposes an interoperability mandate, CEA calls for the Commission to develop a band plan that will foster interoperability. CEA Comments at 22. Research in Motion similarly calls on the Commission to examine auction design and reevaluate the circumstances that are driving multiple band plans. RIM Comments at As the Commission recognizes, [i]nteroperability has often been important to ensuring rapid and widespread deployment of mobile devices in a new spectrum band. NPRM See, e.g., U.S. Cellular Comments at 23; C Spire Comments at 8-9; CCA Comments at 16; Leap Wireless Comments at 7; MetroPCS Comments at 28; Nat l Telecom. Cooperative Ass n Comments at See CCA Comments at 18; Leap Wireless Comments at 8; C Spire Comments at 5 n

63 which spectrum blocks the nation s largest carriers are (and are not) purchasing, the smaller carriers contend that they could find themselves boxed out of the scale economies that the nation s largest carriers enjoy and unable to secure either the roaming arrangements or the access to first-run device offerings on which success in the consumer marketplace depends. 196 As Leap Wireless explains, because small, midsize, and regional carriers rely on roaming arrangements and device ecosystems largely controlled by larger competitors, they have a particular need to evaluate the bids of such competitors, including bids in adjacent markets, to accurately assess the value of particular licenses. 197 Rather than eliminating anonymous bidding and the many administrative, revenue, and fairness benefits it can achieve, 198 the Commission can prevent the anticompetitive consequences of deterministic block assignments that the smaller carriers have identified by simply assigning blocks in a quasi-random fashion across MEAs. With random assignments, competitive carriers are assured that all winners will be assigned a random block. As a result, even the carriers with limited holdings will know that some of their blocks fall within the same frequency as those of the two largest carriers. Thus, the largest carriers could not focus their purchasing power on some blocks in ways that foreclose the smaller carriers from scale economies, raise the costs of purchasing devices, delay access to first-run devices, and preclude or obviate the need for mutually beneficial roaming agreements CCA Comments at 18. CCA asserts that anonymous or blind bidding in Auction 73 resulted in the larger auction participants herding the smaller carriers into the Lower 700 MHz A Block. Id. 197 Leap Wireless Comments at See Auction of AWS-1 and Broadband PCS Licenses Rescheduled for August 13, 2008, Public Notice, Docket No , 23 FCC Rcd 7496, (May 16, 2008) (recognizing that anonymous bidding has competitive benefits, including reduced opportunities for bid signaling, retaliatory bidding, and other anti-competitive strategic bidding). 199 The PCS block, for example, has never experienced an interoperability crisis in large part because no dominant operator has ever consolidated a spectrum position on a single block. No operator could risk developing a boutique 55

64 3. Making the Clearing Rule Easier to Satisfy. Employing a random or quasi-random assignment process keeps bidding activity focused on clearing spectrum. 200 Bidders facing two auctions rather than one will reduce their primary forward auction bids based on their expected activity in the follow-on auction. Reduced primary auction bids will decrease the amounts offered to broadcasters to relinquish spectrum, which, in turn, will decrease the amount of spectrum to be cleared for mobile broadband. Precisely how the Commission could prevent bidders from responding in ways that would reduce the amount of cleared spectrum without a random assignment process is not clear. The Commission will need to determine whether or not the clearing rule is satisfied before the assignment stage because the auctioneer would need to reveal information about the distribution of wins to allow assignment-round bidders to know where they stand. Revealing information about the winners before the clearing rule is satisfied risks revealing critical information that could damage the forward auction bidding. For example, if there were seven licenses available in an MEA and a bidder is winning four licenses and can only choose between licenses one through four or four through seven, then this bidder could infer that in this area there is exactly one more winner of three licenses. Knowing where rival bidders stand provides important information that, if the primary forward auction has not yet closed, the bidder could use to its advantage by, in this example, acquiring a fifth license in that MEA and frustrating the rival s intent to acquire three licenses in the same area or by trying to split licenses at three each at a low price once the bidder knows there is only one other competitor in the area. Random or quasiband class for a single frequency because no one operator holds a single frequency block across a majority of geographic areas. 200 AT&T Comments at 42; see also T-Mobile Comments

65 random assignments reduce the risk of strategic behavior that could frustrate a successful forward auction. 4. Raising More Revenue. The random-rule assignment can also generate more revenue in the forward auction than would be the case if the Commission employed a follow-on assignment auction. Suppose that there are two licenses in an area and three bidders, each demanding one license. Suppose further that License A is 10% better than license B because, for example, it experiences less interference from adjacent TV stations. Finally, suppose that Bidder 1 values license B at 140; Bidder 2 values it at 120; and Bidder 3 values it at 100. Regardless of the license assignment process, Bidder 3 will set the price of the license because Bidder 3 will bid up to its value of winning a generic license. In the case of a random-order assignment, Bidder 3 would bid up to 105 because if he won a license, he would have equal probability of getting the more valuable License A, and the average value of the A and B license combined is 105 (( )/2=105). If Bidder 3 expected an assignment auction, however, he would expect not to be able to outbid the other winner for License A and hence Bidder B would drop out of the forward auction at 100, resulting in forward auction revenue that is lower by ten than under random assignment. In this way, using a random assignment process rather than a follow-on assignment auction can cause the forward auction to raise more revenue for broadcast clearing, public safety, and deficit reduction While the highest price Bidder 1 is willing to pay in this example is lower under the random assignment process, that fact does not reduce revenues because Bidder 1 does not affect the price in the forward auction. 57

66 C. The Commission Should License the 600 MHz Spectrum By Major Economic Area. Licensing the 600 MHz primarily on an MEA basis with the possibility of some EA licenses would (1) satisfy the needs of most wireless carriers; (2) preserve the ability of licensees that wish to provide service in a smaller geographic area; and (3) reduce the need to permit package bidding in an already complex auction. The exclusive use of geographic areas smaller than MEAs creates serious exposure risk for bidders that wish to provide national or regional service and increases the need for a package bidding process. Exposure risk creates a fundamental problem for companies such as T-Mobile that seek to challenge the two largest carriers in gaining access to the 600 MHz band spectrum. To compete on the national level using 600 MHz spectrum, T-Mobile must acquire 600 MHz spectrum virtually everywhere. If 600 MHz licenses were sold only on an EA basis, T-Mobile might have to spend billions of dollars before learning that the total price for the bundle of licenses T-Mobile wants makes the whole entry unaffordable or unprofitable. 202 Without some mechanism to aggregate an economically efficient minimum set of licenses, companies winnings could fall short of what is needed to compete effectively in the 600 MHz band and the risk of such an outcome could compel T-Mobile to sell its 600 MHz licenses to the two largest carriers at firesale prices rather than pursue a competitive network deployment that has been rendered uneconomical for want of sufficient scale or sufficient geographic scope Jeremy Bulow, Jonathan Levin, and Paul Milgrom, Winning Play in Spectrum Auctions, (February 2009), Id. 58

67 Several commenters agree that EA licenses pose too great an exposure risk to encourage robust bidding. 204 Faced with the risk of acquiring markets that lack sufficient scale, potential bidders could simply forego participation in the forward auction, which could hinder the development of competitive wireless broadband services by keeping valuable spectrum from the wireless market. The two largest carriers support limited package bidding with pre-defined packages based on Major Economic Areas (MEAs), Regional Economic Area Groupings (REAGs), or even nationwide licenses. Given its complexity in this context, however, package bidding will prove difficult. The far simpler alternative is to adopt a larger geographic license area, namely the MEA, as the primary license area. Large geographic area licenses such as the MEA reduce exposure risk and allow most carriers to more readily assemble a footprint commensurate with the market they seek to serve. Some commenters, however, remain unconvinced that small EA licenses will expose national carriers to serious financial losses that risk diminishing auction participation and dampening bidding activity. 205 But simply because smaller bidders favor small licenses does not mean that using them would be efficient or improve the consumer experience. Instead, wireless consumers have spoken: they want contiguous service over large areas. U.S. Cellular, however, contends that the use of license areas as small as CMAs or even smaller will not disadvantage competitive carriers that seek to acquire a national or regional footprint. 206 U.S. Cellular even claims that T- Mobile s success in acquiring a nationwide footprint during the AWS auction offers as an example of a carrier successfully assembling a nationwide footprint with smaller geographic area 204 See, e.g., Verizon Comments at 49; AT&T Comments at 52 & Exhibit B. 205 See, e.g., U.S. Cellular Comments at 10-11; CCA Comments at 14-15; MetroPCS Comments at U.S. Cellular Comments at

68 licenses. 207 U.S. Cellular is mistaken. T-Mobile acquired its national AWS footprint using large, REAG geographic areas, not small EAs or CMAs. 208 Moreover, while smaller carriers can acquire a footprint larger than their preferred market area by forming bidding consortia with other smaller carriers and then partitioning the spectrum, carriers with a larger geographic footprint have no such alternative: absent larger geographic areas or package bidding, they have no means of avoiding the exposure risk created by small geographic area licenses. While the differences between smaller carriers that want smaller licenses and larger carriers that want larger licenses are not readily reconciled, a compromise may be possible: the Commission could use MEAs as the base geographic unit but disaggregate those MEAs that have portions of their territory encumbered by broadcast uses into their component EAs. This approach would provide MEA licenses to minimize exposure risk and, in encumbered areas, offer EA licenses as a means of promoting entry by smaller carriers and raising spectrum clearing targets that might otherwise be reduced over a large area. 209 Deconstructing the New 207 See id. at T-Mobile acquired a nationwide footprint during the AWS auction by purchasing larger geographic area licenses, not by assembling smaller ones. The AWS auction involved six different spectrum blocks comprised of geographic area licenses of different sizes: Block A was comprised of 734 Cellular Market Area (CMA); Blocks B and C were each comprised of 176 Economic Area (EA) licenses; and Blocks D, E, and F were each comprised of twelve large Regional Economic Area Groupings (REAGs). In T-Mobile s case, the vast majority of the company s AWS spectrum holdings and the foundation of the nationwide AWS footprint T-Mobile holds rests not on the smallest CMA-sized area licenses of Block A or even the small EA-sized areas of Blocks B and C, but rather on the large REAG-sized area licenses of Blocks E and F. Far from supporting the notion that CMA or EA geographic-area licenses can allow the assembly of a nationwide footprint, T-Mobile s experience in the AWS auction reinforces the need for larger geographic area licenses. While SpectrumCo acquired a national footprint using EAs, this result is generally regarded as an aberration the result of savvy strategic bidding, circumstances peculiar to the AWS auction (in particular, that both small and large licenses were offered nationwide and the activity rules that prevented arbitrage between them), and more than a fair amount of luck. See Jeremy Bulow, Jonathan Levin, and Paul Milgrom, Winning Play in Spectrum Auctions, 1-3 (February 2009), In an incentive auction, the amount of spectrum available for broadband use will depend upon the amount of spectrum the reverse auction can clear. If too many broadcast incumbents remain in a portion of the geographic area, then the entire geographic area would offer less spectrum for broadband use. Suppose, for example, the reverse auction in the New York City portion of the New York MEA is very successful and clears 120 megahertz of spectrum, but the reverse auction in upstate New York is considerably less successful and clears only 60 megahertz 60

69 York MEA into its composite EAs, for example, might allow higher levels of spectrum clearing in New York City EA than otherwise possible if cross-border interference issues in the Vermont, New Hampshire, and upstate New York EAs constrained the amount of spectrum-clearing that could occur throughout the entire MEA. Under this construct, MEAs would remain the base licensing unit and limit the exposure risk, but the selective disaggregation of EAs for all blocks would allow a measure of flexibility and increase the total amount of spectrum clearing. Employing a mix of licenses in this manner can function well even in an auction comprised of otherwise roughly fungible spectrum blocks. Suppose, for example, an MEA has three EAs, called EA 1, EA 2, EA 3. Suppose further that, in a 35x35 MHz band plan, the reverse auction clears seven paired licenses nationwide or MEA-wide with variable amounts of supplemental downlink in each of the three composite EAs. In EA 1, there is one supplemental downlink possible; in EA 2 there are two supplemental downlinks possible; and in EA 3 there are three supplemental downlinks possible. In this example, the nested MEA approach would rely on MEAs for the seven paired blocks and for the one supplemental downlink common to each EA 1 EA 2 EA 3 MHz cleared Paired SDL of spectrum. If the only geographic unit available is the MEA, then the total amount of spectrum available at auction in the entire MEA would be no more than the 60 megahertz available in the most constrained portion of the MEA. In an incentive auction, smaller geographic areas work as a kind of quarantine on encumbrances. The smaller the geographic area is, the lower the risk is that an encumbrance will infect other areas and force lower spectrum-clearing targets across a large swath of otherwise available territory. 61

70 of the three composite EAs within the MEA. But the Commission would create an additional supplemental downlink block on an EA basis in EA 2 and two additional supplemental downlink blocks on an EA basis in EA 3. This nested MEA approach maximizes the spectrum available for licensing while still recognizing and placing a priority on minimizing the exposure risk that those auction participants most apt to participate aggressively in the 600 MHz auction are likely to face with smaller geographic licensing units. This approach would also resolve the overflow problem AT&T identified without the additional complexity of package bidding or the cost of losing some cleared spectrum. 210 D. The Commission Should Allow Package Bidding for Geographic Areas if It Adopts EAs, for Unpaired Blocks, and, if It Adopts Safeguards Against Abuse, for Paired Blocks. While geographic package bidding is complex in the context of an incentive auction, national carriers need some mechanism to mitigate the risk that they will acquire some, but not all, of the licenses they need to create a national footprint. If MEAs are not adopted, the Commission should allow package bids for combinations of spectrum in multiple geographic areas. Allowing additional package bidding for less complex combinations of paired and unpaired spectrum within a geographic area and, under certain circumstances, combinations of paired spectrum within a geographic area would also limit exposure for carriers. Because permitting package bidding for combinations of paired spectrum within a geographic area poses substantial risks of inordinate concentration of spectrum resources, however, the Commission should not permit package bidding for paired spectrum unless the Commission also adopts clear competitive safeguards against undue spectrum concentration below 1 GHz. 210 AT&T Comments, Exhibit B, Yeon-Koo Che, Phil Haile, and Michael Kearns, Design of the FCC Incentive Auctions

71 Geographic Package Bidding. Licensing by MEA would reduce the need for package bidding in many cases because a single license covering an MEA would be sufficiently large to provide a commercially viable service. If the Commission does not adopt MEA geographic license areas to minimize the financial risks that carriers will face in acquiring a national footprint in the 600 MHz band, then it should adopt package bidding for multiple EA geographic area licenses. As the Commission noted, [p]ackage bidding could be particularly helpful to bidders that face a risk of winning certain licenses but losing complementary licenses they consider essential to their business plans. 211 In the absence of larger MEA licenses, allowing bidders the opportunity to make package bids that are contingent on obtaining other spectrum units would encourage broader auction participation, increase auction revenues, and enable greater efficiencies for carriers seeking to deploy 600 MHz service across a wide footprint. 212 A national carrier such as T-Mobile needs a national or nearly national footprint before investing the considerable funds needed to develop, deploy, and support a new spectrum band for its existing and prospective customers. Paired-Unpaired Package Bidding. Failing to win paired 600 MHz spectrum could make deployment of unpaired 600 MHz spectrum uneconomical. Without paired spectrum, the per unit costs of deploying the necessary equipment to support downlink-only 600 MHz operations at the network and handset level is made more expensive. Whereas a bidder with 211 NPRM See Spectrum Exchange Group, LLC, FAQs about Ascending Auctions with Package Bidding 1 (2000) (explaining that the exposure problem occurs when a bidder faces the risk of acquiring only some of the licenses that are necessary to carry out its business plan. If the bidder fails to acquire some of the licenses it needs, the complimentary licenses the bidder acquires are not worth the prices paid.); see also Christoph Brunner et al., An Experimental Test of Flexible Combinatorial Spectrum Auction Formats 2 (2007) (describing how package bidding eliminates the exposure problem by allowing bidders to submit bids that include combinations of complementary licenses and allow the bidder to either win the entire package or nothing at all, and as a result, bids can reflect value complementarities, which should raise efficiency and seller revenue ). 63

72 paired spectrum could rely on the same device and network hardware (such as antennas, filters, and similar frequency-related components) to support downlink-only operations in proximate bands, a bidder without paired spectrum in the vicinity would have to make difficult decisions about whether or not to invest the developmental, financial and physical resources to support an additional band not already in its portfolio. If the winning bidder has not already committed to develop the hardware necessary to deploy paired 600 MHz spectrum, holding a small outpost of downlink-only spectrum in a distant band may simply not prove worth the carrier s time and money to develop. The exposure risk that results from a lack of package bidding with paired and unpaired spectrum will prove especially problematic for T-Mobile, which is unique among the four nationwide carriers in having virtually no low-frequency spectrum that unpaired 600 MHz could readily and logically supplement. Allowing for package bidding for paired and unpaired blocks in the 600 MHz band would greatly reduce the uncertainty for T-Mobile and other bidders that lack access to low-frequency spectrum resources and minimize the risk that bidders will wind up with licenses they cannot economically use. One potentially attractive alternative to offering package bidding for paired and unpaired spectrum is to permit limited bid withdrawals for unpaired spectrum bids. A bidder that is outbid on paired spectrum might still hold the top bid on unpaired spectrum. Rather than leaving these bidders stranded with unpaired spectrum that may prove uneconomic to deploy, the Commission could allow them to withdraw their bids on the unpaired spectrum. Bidders can use bid withdrawals strategically, and the Commission would have to adopt a few clear limitations to avoid gaming the bid-withdrawal process. First, because the bid withdrawal is designed to prevent a bidder on paired and unpaired spectrum from becoming stranded with only unpaired spectrum, the bidder would have had to have bid on paired spectrum to become eligible to 64

73 withdraw its bid on unpaired spectrum. Second, because the exposure risk to unpaired spectrum results from economies of scale in equipment procurement, the bidder would have to have bid on some substantial amount of both paired and unpaired spectrum. Third and finally, because multiple withdrawals could be used strategically during the auction regardless of the other safeguards that may be in place, bidders who are eligible for withdrawal should be limited to only one withdrawal of a bid on unpaired spectrum for lack of complementary paired spectrum during the incentive auction. Paired Spectrum Package Bidding. While T-Mobile supports the Commission s proposal to auction the 600 MHz spectrum in 5x5 MHz spectrum blocks, this amount of paired spectrum falls short of both the increasingly large blocks of contiguous spectrum contemplated for LTE and the 10x10 MHz blocks already deployed or under deployment by the two leading national carriers. Especially for carriers such as T-Mobile that hold no low-frequency spectrum resources, the 600 MHz auction represents an important opportunity to provide consumers with a high-performance broadband network that can readily penetrate inside buildings and propagate over large geographic areas with minimal infrastructure. Acquiring low-frequency resources in combinations of 10x10 MHz or greater would enable T-Mobile and other competitive carriers to develop low-frequency, coverage networks and potentially allow them to reduce the amount of roaming fees they must pay the two largest wireless providers that hold most of the lowfrequency spectrum in the U.S. market today. Yet, without some limit on the two largest carriers ability to package paired spectrum in the 600 MHz auction, the risk that they would pay a premium to exclude carriers such as T- Mobile from these valuable, low-frequency resources is great. While the two largest wireless providers may also achieve some potential synergies between the 600 MHz band spectrum and 65

74 existing spectrum and infrastructure resources, they will have a strong incentive to deter competitive entry in the low-frequency spectrum bands that they dominate today. Absent a limit on spectrum holdings below 1 GHz, the market concentration resulting from allowing the two largest wireless providers to place package bids on paired spectrum would harm consumers and reduce total economic benefit. 213 Unless the Commission establishes safeguards to prevent concentration of spectrum holdings below 1 GHz, the costs to competition of allowing this type of package bidding would likely outweigh the potential incremental benefit of allowing package bidding for paired blocks in the 600 MHz band. E. Without Careful Oversight and Stiff Penalties for Abuse, Special Bidding Credits Will Harm the Auction and Damage Competition. Bidding credits create powerful incentives to acquire wireless broadband spectrum. 214 These subsidies not only provide a strong inducement for market entry, but also can entice applicants to try to capture a portion of the subsidy for themselves. As one economist noted, bidding credits encourage companies to create eligible bidders that are carefully constructed to satisfy the rules but circumvent their intent. 215 Even the most conscientious recipient of bidding credits still faces the prospect of a large payout if they can lease or sell the spectrum or spectrum capacity that, by design, is worth more than they actually paid Peter Cramton et al., Using Spectrum Auctions to Enhance Competition in Wireless Services 5 (2011), available at See Implementation of the Commercial Spectrum Enhancement Act and Modernization of the Commission s Competitive Bidding Rules and Procedures, Second Report and Order and Second Further Notice of Proposed Rulemaking, Docket No , 31 FCC Rcd (Apr. 25, 2006) ( Bidding Rules Modernization 2d R&O and 2d FNPRM ). 215 Peter Cramton, Professor of Economics, University of Maryland, Lessons from the United States Spectrum Auctions, Prepared Testimony before the Senate Budget Committee (Feb. 10, 2000). 216 Peter Cramton et. al, Using Spectrum Auctions to Enhance Competition in Wireless Services, 54 Chi. J. L. & Econ. S167, S176 (2011), 66

75 To combat these incentives, the Commission has had to expend significant time and energy in monitoring compliance with bidding credit rules. The anti-windfall rules have included requiring entities to seek advance approval for sales, applying additional scrutiny to deal by eligible entities, and randomly auditing recipients of eligible entities use of bidding credits. 217 Even then, ensuring compliance has proven challenging. 218 As the Commission has recognized in policing bidding credits, companies strategies continually evolve to take advantage of the bidding credit payout. 219 If bidding credits are adopted, therefore, the Commission must adopt detailed eligibility criteria, exhaustive limitations on flipping, and robust compliance audits. The Commission must also swiftly apply meaningful sanctions in the event of non-compliance. To preserve competition, the Commission must also ensure that eligible entities are not used by the dominant carriers to skirt whatever spectrum caps or screens the Commission may adopt in this proceeding or elsewhere. Finally, the spectrum acquired using those credits should be fully attributed to the party who actually uses it whether that spectrum or spectrum capacity is acquired through lease, option, joint venture or any other form of ownership or collaboration, broadly defined. 217 Bidding Rules Modernization 2d R&O and 2d FNPRM 42 (explaining that it has witnessed a growing number of complex agreements between entities receiving bidding credits and those with whom they choose to enter into financial and operational relationships ). 218 Peter Cramton, Professor of Economics, University of Maryland, Lessons from the United States Spectrum Auctions, Prepared Testimony before the Senate Budget Committee (Feb. 10, 2000) ( In our opinion the use of bidding credits for small businesses in the US spectrum auctions did not have a major impact on post-auction competition. ). 219 Bidding Rules Modernization 2d R&O and 2d FNPRM

76 IV. THE REVERSE AUCTION A. Addressing the Most Important Reverse Auction Concerns As Soon As Possible Would Promote Broadcaster Participation and Accelerate the Incentive Auction. The Commission has already acknowledged the importance of generating broad participation in the reverse auction. 220 Meeting this goal means creating a simple and transparent process for broadcast licensees to submit bids to voluntarily relinquish or share their spectrum. 221 To this end, the Commission has engaged in a number of outreach efforts to inform the broadcast community of the issues presented in this proceeding, and intends to continue those efforts in the coming months and years. 222 As Commissioner Pai recently noted, [t]he broadcast incentive auction is inherently complicated and unnecessary complexities are likely to deter participation. 223 Chairman Genachowski has likewise expressed the Commission s commitment to providing broadcasters the information necessary for them to make sound business decisions and to help [broadcasters] recognize the full value of the opportunity afforded by this auction. 224 One of the principal means by which the Commission can deliver certainty to the market and increase participation in the reverse auction is by resolving as many reverse auction issues as possible, as soon as possible. By promptly establishing the regulatory framework for spectrumclearing, channel reassignment, reverse bid options, and radiofrequency interference criteria among other issues the Commission will enable broadcasters to determine whether the reverse 220 See NPRM See id. 222 Id. 223 Statement of Ajit Pai, Federal Communications Commission Hearing Before the Subcommittee on Communications and Technology of the United States House of Representatives Committee on Energy and Commerce (Dec. 12, 2012). 224 Remarks of Julius Genachowski, National Association of Broadcasters Show 2012 (Apr. 16, 2012). 68

77 auction offers a sound business opportunity for them, and provide them with ample time to identify viable business models, investor preferences, and strategic plans. 225 In addition, as noted by Verizon, by providing a known spectrum-clearing target, the Commission could provide interested parties with a better understanding of where broadcast and mobile operations are likely to occur in the 600 MHz band after the auction. 226 In markets where the repacking formula demonstrates that the Commission can achieve the spectrum-clearing target completely through repacking, broadcasters would know in advance that their stations in those markets will not be part of the reverse auction, and could plan accordingly. Likewise, this information would assist broadcasters in planning their auction strategy in those markets where repacking would not alone clear sufficient spectrum. 227 The resulting certainty would also benefit new 600 MHz licensees required to deploy their wireless facilities in a timely manner, in accordance with any performance standards that the Commission adopts. To address the critical issues regarding the reverse auction, the Commission anticipates issuing a series of public notices in the future that will provide additional opportunities for interested parties to comment on incentive auction design issues. 228 However, as noted by Verizon, deferring these matters, rather than expediting their resolution, could frustrate the Commission s objectives in this proceeding, delay the completion of the 600 MHz reallocation, and reduce participation among interested parties. 229 Given the complexity of this proceeding, T-Mobile joins Verizon in urging the Commission to adopt rules concerning the reverse auction as soon as possible, even if other issues remain pending or warrant further deliberations. 225 See Verizon Comments at Id. at See Verizon Comments at NPRM See Verizon Comments at

78 Expediting the most salient reverse auction concerns should be paramount if the Commission intends to begin this proceeding in B. Using Multiple Opening-Round Bids Promises a Simpler and Faster Auction that is More Likely to Satisfy the Clearing Rule than Other Alternatives. Offering broadcasters multiple opening bids will make the auction simpler to conduct and faster to complete. The incentive auction requires the Commission to make opening bids for broadcast stations throughout the country, and T-Mobile has generally encouraged the Commission to start high. 230 As a rule, competition among reverse auction participations should ensure that the price for broadcast stations remains reasonably related to the prices forward auction bidders prove willing to pay. In some markets, however, an insufficient number of broadcasters may participate in the incentive auction to ensure pricing that accurately reflects the willingness of the marginal broadcaster to sell its spectrum at a given target level of clearing. This phenomenon could needlessly increase the cost of clearing the broadcast spectrum and, perhaps more importantly, could artificially reduce the total amount of spectrum cleared. For example, the Commission may set a spectrum-clearing target that involves four broadcast stations in a region. If four broadcasters are willing to cease broadcasting at a price of $100, but the fifth broadcaster is only willing relocate at a price of $200, no competitive force would exist to drive the price below $200. Such a large gap between private value and collective payment creates substantial inefficiencies and could cause the auction to fail to satisfy the minimum revenue clearing rule. 230 T-Mobile Comments at 46; TIA Comments at 13; Comments of the Expanding Opportunities for Broadcasters Coalition, GN Docket No , at 10 (Jan. 23, 2013) ( EOBC Comments ); Comments of a Prospective Reverse Auction Participant, GN Docket No , at 6-7 (Jan. 23, 2013). 70

79 One suggestion to prevent market failure is the use of reference prices for broadcast stations from other areas. 231 While this idea of using prices from other areas as a surrogate for in-market competition has some potential, it remains unclear how the Commission might incorporate this concept into a voluntary spectrum auction without risking missing the clearing target. That is, the reference price could lead a broadcaster to believe that its participation is required for the Commission to meet its spectrum-clearing target and refuse to exit the band at a low reference price. As an alternative (or a complement) to pursuing a more detailed and predictable means of establishing reference prices for markets across the country, the Commission should simply offer broadcasters multiple opening bids. 232 This process is straightforward for the Commission to administer and for broadcasters to understand. Rather than only ask a broadcaster whether it would leave the 600 MHz band for $200, for example, the Commission would ask whether the broadcaster would leave the 600 MHz band for $240, $190, and $140. In response, the broadcaster would simply need to provide a series of yes or no answers, such as yes to $240, yes to $190, and no to $140. In this way, the Commission could set a very high initial price and additional lower prices. Each broadcaster would indicate whether it would be willing to cease broadcasting at each of the opening prices. In essence, this process would create a simple sealed bid for broadcasters in the initial round of bidding. The Commission would determine the lowest price where there were enough broadcasters to satisfy the target clearing and then start the descending clock auction at that price. In situations where more broadcasters were willing to cease operations at the lowest price than there was demand for the spectrum, the clock would 231 See Auctionomics-Power Auctions, Option for Forward Auction 4-5 (Feb. 1, 2013) ( APA Joint Filing ), available at T-Mobile Comments at

80 decrease. In cases where the number of broadcasters was exactly equal to the spectrum-clearing target, the clock would not decrease. Using the example above with opening prices of $240, $190 and $140, each of the four broadcasters and the fifth broadcaster would decide which, if any, of the Commission's three offers it would accept. With a single high opening bid of, say, $240, the clock would decrease to $200 and stop there. With multiple opening bids, all four low-value broadcasters might signal a willingness to accept $190 or even $140, which would then be the price in the reverse auction. This process would make the incentive auction more likely to satisfy the clearing rule while enhancing efficiency. In economic terms, using multiple opening bids provides a very simple mechanism to incorporate an element of a first-price auction into the second-price general design, which is precisely the approach leading theorists have advocated for auctions that incorporate a risk of low competition. 233 C. Sequentially Alternating the Reverse and Forward Auctions Balances the Need for Simplicity and Speed Against the Desire to Allow Participants to Monitor and Respond to a Changing Auction Environment. Commenters support an alternating or staged approach to the reverse and forward auctions. 234 As Verizon explains, under such an approach, the Commission would commence the auction process by conducting an individual stage of the reverse auction. 235 The Commission would collect reverse auction bids based on a target amount of broadcast spectrum to be cleared for the initial round, an elevated best-case scenario target. 236 The Commission would then 233 See, e.g., Paul Klemperer, What Really Matters in Auction Design, 16 J. Econ. Persp. 169, (2002) (explaining that a hybrid Anglo-Dutch auction, which combines elements of ascending and sealed-bid auctions, encourages competition where competition is otherwise expected to be weak), available at T-Mobile Comments at 42-44; Verizon Comments at See Verizon Comments at Id. 72

81 conduct the forward auction, with prospective wireless licensees bidding on the wireless licenses that would be created if the target from the preceding reverse auction stage is cleared. 237 After a single stage of each of the reverse and forward auctions is complete, the Commission would then determine if the forward auction bids are sufficient to cover the reverse auction bids and meet the overall auction closing conditions. 238 If the total forward auction bids are sufficient, the incentive auction would be closed and the process complete. If, however, the aggregate forward auction bids fall short of the closing conditions, the Commission would try to coax additional bids out of the forward auction participants to meet the closing conditions. 239 And if the total forward auction bids still fall short after attempts to coax higher bids, then the Commission would start a new round of reverse and forward auction stages by incrementally reducing the spectrum target for the new stage of the reverse auction. 240 This staged process of alternating reverse and forward auctions would continue with incremental reductions to the spectrum target until the closing conditions of the auction are met. 241 Important common value considerations among the broadcasters may exist that would prompt broadcasters to update their valuations based on the results of the forward auction. For example, if the forward auction fails to clear a 120 megahertz target, the size of the shortfall may provide important information to the participating broadcasters about how much the forward- 237 Id. 238 See NPRM See infra Part IV. C. 240 Verizon Comments at Verizon s proposed staged approach for the forward and reverse auction is largely similar to T-Mobile s proposed alternating approach. T-Mobile Comments at 42 n.92. The only notable difference is that T-Mobile suggested that the Commission consider conducting multiple stages of the reverse auction, by seeking bids at more than one clearing target, before switching to the forward auction. Id. To the extent that conducting multiple stages of the reverse auction before alternating to the forward auction would speed and simplify the auction process, the Commission should consider it as an option for conducting the staged auction. 73

82 auction participants value their spectrum and may cause some broadcasters to change the value they have placed on retaining a license. The logical alternative to sequentially alternating the forward and reverse auctions is to conduct a one-iteration reverse auction in which the Commission would run the reverse auction from the highest possible price with the most clearing all the way down to whatever price resulted in no broadcast incumbents exiting the band. This process would provide the Commission with considerable information about the price at which the auction could satisfy the full range of spectrum-clearing targets in every geographic area in the country. Unlike the single-round, sealed-bid approach a one-iteration reverse auction would permit the spectrum incumbents to learn from one another as the auction progressed, which would steadily lead to better, more informed pricing across a wide range of values and geographies. Unfortunately, however, a one-iteration reverse auction would not provide any feedback to the broadcasters about the buyers demand for broadcast spectrum licenses. A one-iteration reverse auction would also have to run its entire course before the forward auction could begin and, as a result, would require considerable time to complete and likely create major business, financial, logistical and legal challenges for participating incumbents. 242 An alternating auction format, by comparison, would provide much of the same helpful information in far less time. Although buyers would not develop information about the full supply curve, they would develop information about relevant portions of it during the alternating sequences. The sellers could also monitor forward-auction bids to gauge whether the bidders seemed likely to prove able to meet or exceed the broadcaster s value assessment. In this way, alternating forward and reverse 242 See, e.g., Verizon Comments at 26 (explaining that broadcasters must reveal more sensitive information under a one-iteration reverse auction approach, including the offer price for their channel(s) under multiple clearing targets). 74

83 auction participation would allow the buyers and sellers to benefit from the participants on both sides of the market, which would help refine their estimates of the common value of the spectrum resource. D. Offering Broadcasters Too Many Exit Options in the Reverse Auction Could Produce Excessive Uncertainty and Delay or Disrupt the Auction Process. The costs of offering reverse-auction participants options not required by the Spectrum Act outweigh the benefits. The Spectrum Act requires the reverse auction to include three bid options for participating broadcasters: (i) voluntary relinquishment of all spectrum usage rights, (ii) voluntary relinquishment of all UHF usage rights in exchange for VHF spectrum rights, and (iii) voluntary relinquishment of spectrum usage rights in order to share a television channel with another licensee. 243 In the Notice, the Commission sought comment on whether to allow eligible reverse auction bidders to submit other types of bids, including bids to accept additional interference, reduce coverage areas, or accept a different antenna pattern. 244 The Notice also stated that one of its goals is to permit as many broadcasters to participate in the reverse auction as possible, and explained that allowing reverse auction bidders additional bid options could enable the Commission to clear more spectrum. 245 At the same time, the Commission expressed concern that allowing too many challenging or variable bidding options might also significantly complicate the reverse auction process. 246 While adding options for broadcasters has the potential to reduce the cost of clearing the spectrum, the Commission s concerns about complexity are well founded. T-Mobile agrees with 243 Spectrum Act 6403(a)(2). 244 NPRM Id. 72, Id

84 commenters who have expressed concern that offering broadcasters too many additional bid options beyond the statutory minimum risks unduly complicating the reverse auction, and that any benefits would be outweighed by additional uncertainty, delays, or disruptions to the auction process. 247 Offering an array of variable, fact-specific options to exit the 600 MHz band creates excessive complexity that risks diminishing broadcaster participation precisely the opposite effect of what the Spectrum Act intended and what the Commission seeks to achieve by providing alternative bid options. Keeping the auction design simple is particularly important to encourage broadcaster participation because the auction process is entirely new to many of the broadcast licensees. 248 Moreover, [m]aking the process as straightforward and simple as possible will foster participation by smaller broadcasters who may be unwilling to put together an elaborate team of auction experts to help them navigate a complicated process. 249 Furthermore, overly complicating the reverse auction could have negative consequences for the forward auction because if broadcasters are able to bid to accept additional interference or reduce their coverage areas rather than vacate UHF television channels, forward-auction participants may find it difficult to understand what they are actually bidding for. 250 Even proponents of alternative mechanisms to exit the 600 MHz band acknowledge that additional bid options could complicate the reverse auction and discourage participation. 251 CTIA, for example, urged the Commission to strike an appropriate balance between simplicity 247 See e.g., EOBC Comments at (explaining that the Spectrum Act specifically contemplates three bid options for broadcasters, and in doing so Congress appropriately balanced the utility of providing flexible bid options in furtherance of its goal of reallocating at least 120 MHz for mobile broadband use against the risk of over contemplating the reverse auction and undermining those efforts ); MetroPCS Comments at 5 (noting the auction should be designed to foster wide participation by broadcasters, but also avoid unnecessary complexity); Sprint Comments at 6; T-Mobile Comments at MetroPCS Comments at Id. 250 Sprint Comments at See e.g., CEA Comments at 13; CTIA Comments at 33; TIA Comments at 14-15; Verizon Comments at

85 and ensuring maximum flexibility for broadcasters. 252 While TIA praised the potential of additional bid options, it noted that [a]n overly complex... decision tree may be intimidating for some broadcasters, particularly those with limited resources to invest in educating themselves about the reverse auction process creating the possibility that a complicated auction may discourage some TV licensees from participating. 253 And while Verizon favored the potential benefits of providing broadcasters additional flexibility, it observed that [a]llowing broadcasters to submit an alternative bid... would likely add a degree of complexity to the Commission s administration of the auction, including the bid assignment and repacking methodology. 254 Only a handful of commenters, including some broadcasters and equipment manufacturers, support allowing alternative bids without reservation. 255 And notably, those who advocate for additional bidding options do not address how the Commission or the bidders would resolve the additional complexity that these alternatives would introduce. While T-Mobile strongly supports any measure that clears more valuable below 1 GHz spectrum for broadband use, the risk that offering too many exit options will diminish participation and delay the incentive auction outweighs the potential benefit of clearing additional spectrum in this case. Allowing broadcasters to submit bids to accept additional interference, reduced coverage, or different antenna patterns would create needless challenges for reverse-auction and forward-auction participants as well as the Commission itself. In short, the benefits of pursuing alternative bidding options do not outweigh the costs. 252 CTIA Comments at TIA Comments at Verizon Comments at See Comments of Harris Corporation, Broadcast Communications Division, GN Docket No , at 23 (Jan. 25, 2013) ( Harris Corp. Comments ); Qualcomm Comments at 24; Comments of Tribune Company, GN Docket No , at 4 (Jan. 25, 2013) ( Tribune Comments ). 77

86 E. Addressing an Unsatisfied Closing Requirement. Addressing unsatisfied closing requirements by seeking to coax additional money from forward auction participants will help free additional spectrum for wireless broadband. 256 The overall auction ends only if the total bids in the forward auction meet a reserve price for the spectrum. This reserve price, or clearing rule, serves an important function in matching spectrum supply to demand, but poses some strategic problems for forward-auction bidders that could introduce inefficiencies into the auction that suppress efficient spectrum clearing. As T-Mobile and other commenters explained in their initial comment filings, the crux of the problem is that, while every bidder may want additional spectrum, each bidder will always want other bidders to pay to support the higher level of spectrum clearing. 257 Suppose, for example, that there are two licenses, A and B, and three bidders. Bidder 1 values License A at 60; Bidder 2 values License B at 60; and Bidder 3 values each of the two licenses at 50. The clock auction envisioned here would end at a price 50 (when bidder 3 drops out) and Bidders 1 and 2 would win. If the clearing rule requires total revenue of 110, however, then the forward auction revenue in this example would fall short of the requisite target. Even though the winning bidders are willing to pay 120, which would more than satisfy the clearing rule, each bidder has an incentive to delay revealing their preferences in the hope that the other bidder raises its bid enough to meet the shortfall; therefore, the Commission would not award any licenses, but instead reduce the spectrum-clearing target and try again. 256 T-Mobile Comments at 56. AT&T proposes a similar mechanism that would coax additional spectrum clearing from broadcasters who might risk not selling if the auction drops to a low spectrum-clearing target. See AT&T Comments at AT&T s proposal to encourage additional spectrum clearing where the bidders are close to a higher spectrum-clearing target is sensible; however, the free rider problem among reverse auction bidders is liable to prove substantially worse than among forward auction bidders. While the free rider problem may frustrate reverse-auction side measures to satisfy the closing rule, AT&T s insight on this point may merit additional exploration by the Commission. 257 See, e.g., T-Mobile Comments at 56-58; AT&T Comments at

87 The Commission has several means of coaxing additional money from the bidders to clear the maximum amount of spectrum. One approach is to ask each of the forward auction winners to pay its proportionate share of the shortfall. A variant is to avoid a potential free-rider problem with small bidders and allow them to benefit from the propensity of larger bidders to pay more for a higher spectrum-clearing target by allocating the shortfall among bidders with 5% or more of the currently winning bids. Large bidders will likely have at least some licenses with values far above the final prices; therefore, these large bidders are more likely to agree to a lump sum. By comparison, small bidders might prove more reluctant to pay additional amount for the spectrum they seek. Alternatively, the Commission could ask for contributions from only the provisionally winning bidders in areas where the bidder wins more than one paired license on the theory that the extra license that these bidders hold might not be available if the spectrumclearing target were reduced. In either case, the rule would only have a single round where bidders could say yes or no to the call for increased bids. Because each large bidder would know that its participation would be necessary for the higher spectrum clearing-target to be met, bidders would have a reduced incentive to act as free riders on the presumed willingness of other bidders to make up the shortfall. Whether the Commission used a percentage or an absolute number of licenses to identify and seek contributions from larger bidders, this approach would avoid reducing the spectrum-clearing target reduction for all parties by coaxing additional payments from only those forward-auction participants likely to benefit most from additional spectrum clearing and allowing those who benefit less to avoid additional payment Yet another approach would be to stipulate that all bidders applying to participate in the auction would make a deposit proportional to the initial eligibility they sought. In the event of a shortfall between the bids and the spectrum clearing target that would cause the clearing rule to not be satisfied, a portion of the sum of the deposits could be used to cover the shortfall. Under this option, bidders that would be provisional winners in the auction would receive only a partial refund of their deposits. 79

88 Whatever the precise mechanism, adopting a procedure that offers a take-it-or-leave-it option at the end of the auction to help maintain the highest level of spectrum clearing would help mitigate the risk that willing bidders are forgoing additional bids in the mistaken assumption that the other bidders will make up the shortfall. A take-or-leave-it approach at the end of the auction would cut through the gamesmanship and miscommunication that might lead to a lower spectrum clearing target by offering willing bidders a final opportunity to collectively satisfy the clearing rule. Adopting one of the take-it-or-leave-it mechanisms described here would remove much of the uncertainty over how winning bidders would split the shortfall between their existing bids and the minimum price required by the clearing rule and, in so doing, greatly increase the likelihood of a higher spectrum clearing target. F. While the Concept of Extended Rounds Has Great Promise, Modifications Are Needed and a Last-Call Solution May Offer a Simpler Solution to the Problem of a Clearing-Rule Shortfall. Whether through use of the extended rounds proposal or the last-call approach or some hybrid of the two, the Commission should adopt a mechanism to meet any funding shortfall with the closing-rule before reducing the spectrum-clearing targets. Dynamic alternatives to the lastcall mechanism T-Mobile proposed hold promise, but the mechanisms proposed thus far require further refinement. A joint filing by Auctionomics and Power Auctions (the APA Joint Filing ), for example, proposes extended rounds to address a clearing-rule shortfall. 259 The APA Joint Filing is broadly consistent with the demand-matching procedures T-Mobile proposed in its initial comments and, in some cases, incorporates proposals that T-Mobile endorses, such 259 See APA Joint Filing at 6. 80

89 as the concept of gradually reducing spectrum-clearing target, area-by-area, rather than all at once for the whole country. 260 The concept of extended rounds is best explained by example. Suppose the auction begins with a target of eight paired licenses. The forward auction would continue until the demand is no larger than eight licenses. If cumulative bidding satisfied the clearing rule, then the auction would end. If cumulative bidding did not satisfy the clearing rule, however, then the APA Joint Filing proposes a series of extended rounds to bring the auction to a close. 261 Extended rounds would work by reducing the spectrum-clearing target in every area by one license to seven licenses to use the example provided earlier. In any area where the demand supports more than seven licenses, prices would continue to increase. If at any point during the extended rounds the total revenue satisfied the clearing rule, however, the forward auction would stop in all geographic areas. 262 When the auction stops, some areas would have demand sufficient to support clearing eight licenses while other areas would have demand sufficient to support clearing only seven licenses. 263 Although not directly addressed in the APA Joint Filing, the reverse auction would presumably restart in those areas where demand dropped to seven in order to reduce the number of broadcast stations that need to be cleared from that geographic area a process that would reduce the Commission s clearing costs and enhance its net auction revenues Compare id. at 6 with T-Mobile Comments at APA Joint Filing at Id. 263 In all markets where bidding satisfies the clearing rule without demand dropping from eight to seven, the outcome is equivalent to every provider agreeing to pay the lump-sum additional amount to cover the shortfall as T- Mobile envisioned in its last-call proposal. 264 Alternatively, the remaining licenses could be re-auctioned. 81

90 The APA Joint Filing s extended rounds proposal has substantial merit. As with the lastcall model T-Mobile proposed, the extended rounds concept matches forward-auction demand with reverse-auction supply in the event of an incongruity. Unlike T-Mobile s last-call proposal, moreover, the extended rounds concept allows bidders to gather information about the bidding preferences of auction participants and adjust their pricing strategies in response. Despite the dynamic features of the extended round process, certain elements of the proposal appear to rely on assumptions that may result in unexpected outcomes. The APA Joint Proposal, for example, appears to assume that target revenue for the reverse auction will not decrease. However, the total spent in the reverse auction should, in fact, fall with a lower spectrum-clearing target because, even if prices in the reverse auction did not decrease any further, bidders would need to purchase one (or more) fewer television stations than previously. Using the example above of an eight-station market moving to a seven-station clearing target, reverse-auction spending in that geographic area should fall by roughly 1/8 th of the previous bidding total. The decrease in spending associated with a decrease in the spectrum-clearing target would pose a complication for the extended round concept. In a given area, the exact number of television stations needed to achieve the reduced spectrum-clearing target under the extended round model may be unclear due to the daisy chain effect of broadcast stations that have some logical connections to other markets. While the Commission should prove able to develop some mechanism to determine a lower bound on the savings from not having to purchase an additional station in that market, it should not assume it can relax the clearing rule for that market without experiencing a concomitant reduction in the revenue target. The Commission might also consider combining T-Mobile s proposal to offer wireless providers a once-per-target opportunity to cover a clearing rule shortfall on a pro-rata basis with 82

91 the idea of extended rounds. Before starting extended rounds, the Commission might offer participants a last-call opportunity to make additional offers to satisfy the clearing rule. 265 If the last-call method succeeds, the Commission would not have to administer extended rounds. If the Commission settles on a dynamic mechanism such as extended rounds to help satisfy a closing rule over a relative static mechanism such as T-Mobile s last-call proposal, the Commission should incorporate safeguards into the extended rounds to minimize gaming opportunities. For instance, bidders should be prohibited from switching their demands among different areas and instead be allowed only to reduce their demand. For the same reason, any reduction in demand in the extended rounds should count as a reduction of eligibility that counts for the next phase of the forward auction when it is restarted. With these no-substitution constraints, the extended rounds concept is reasonable and should allow bidders to express their preferences in ways that either increase forward-auction revenues, or decrease spectrum-clearing targets in a rational manner. The potential pitfalls associated with the closing rule that the APA Joint Filing identifies pose real concerns that the Commission s auction procedures should address. The APA Joint Filing s proposal provides more opportunities for dynamic responses than T-Mobile s last-call proposal does; however, T-Mobile s proposal may prove easier for the Commission to administer, create fewer opportunities for gaming, and clear more spectrum. On balance, the last-call option seems to offer nearly as many benefits as the extended rounds option but with less administrative complexity and risk. If the Commission nonetheless elects to pursue the extended rounds option, a compromise approach that incorporates some of the elements of T- 265 In principle, that offer could be made every time demand in an area drops by one unit, but it is unclear whether any potential benefits would be worth the increased complexity. 83

92 Mobile s last-call approach, such as multiple spectrum-clearing targets, might provide a superior outcome compared to the original extended round proposal found in the APA Joint Filing. G. Broadcasters Participating in the Incentive Auction Should Not Be Allowed to Revoke Their Bids During or After the Auction. Reverse auction bids should be treated as irrevocable, binding offers of the broadcast spectrum licensee to relinquish spectrum usage rights. 266 A broadcaster s bid will affect the amount of spectrum cleared in a market, which, in turn, will affect the number and value forward-auction bidders place on the spectrum at auction. Allowing a broadcaster to withdraw a bid after the auction is complete would require re-running the auction process for the geographic region where the broadcaster is located and, quite likely, the nation as a whole, given the many different interdependencies that exist between neighboring geographic areas both on the reverseand forward-auction sides. Equally troubling, if a broadcaster were permitted to withdraw a winning bid, then broadcasters would have little or no incentive to bid truthfully. 267 Instead, broadcasters would have a strong incentive to bid below their actual costs for at least two reasons. First, low-ball bids serve as the only reliable mechanism to ensure that the bidder is among the reverse-auction winners in a non-binding auction. 268 Studies show that a bidder who bids his true cost in a nonbinding auction will probably lose. 269 Second, bidders can use low-ball bids in a non-binding auction to punish rivals. A bidder who has no intention of leaving the 600 MHz band, for 266 NPRM 249, See e.g., Brian Merlob et al., The CMS Auction: Experimental Studies of Median-Bid Procurement Auction with Non-Binding Bids (April 2011), available at (discussing the detrimental effects of non-binding bids in the context of auctions operated by Centers for Medicare & Medicaid Services). 268 Id. 269 See id. at 7. 84

93 example, might have an incentive to force prices for broadcast stations down to prevent those broadcasters who are most interested in leaving the 600 MHz band from doing so. 270 Closing off the possibility of a timely and lucrative exit path might make the losing reverse-auction bidders more likely to sell to the untruthful reverse-auction winner. 271 These predictable and unwelcome consequences threaten to unravel the stability necessary for either the reverse auction or the forward auction to function. Furthermore, as Verizon explained, the uncertainty and gamesmanship made possible by non-binding bids would delay forward auction winners in their work to integrate the 600 MHz spectrum into their networks. 272 The Commission therefore should reject the request of a handful of commenters who want to submit non-binding bids. 273 V. RELOCATION AND CLEARING A. The Repacking Process Should Seek to Maximize the Amount of Spectrum Reallocated for Flexible Use. To free additional spectrum for broadband use, the Spectrum Act directs the Commission to repack the broadcast television channels that do not exit the 600 MHz band while mak[ing] all reasonable efforts to preserve... the coverage area and population served of each broadcast television licensee. 274 The Spectrum Act does not require the Commission to preserve the identical service area or population coverage for each station. 275 Nor does the Spectrum Act require a broadcaster s new service area and population to be of the same size or extent as 270 See id. 271 Id. 272 See Verizon Comments at See Tribune Comments at Spectrum Act 6403(b)(1)-(2). 275 Verizon Comments at 36; CEA Comments at 31-32; CTIA Comments at 34-35; AT&T Comments at

94 existed prior to the incentive auction. 276 Instead, the Commission need only make all reasonable efforts to preserve the coverage area and population served of repacked broadcast licensees. 277 Measures that are too costly, too complicated, or too time consuming are inherently unreasonable. 1. The Plain Text of the Spectrum Act Accords the Commission Substantial Flexibility in Repacking Broadcasters. Under the Spectrum Act, the Commission need not achieve perfection in replicating broadcast contours when it repacks and reassigns broadcast channels the Commission need only make all reasonable efforts to preserve... the coverage area and population served of each broadcast licensee. 278 This provision offers the Commission ample flexibility. As AT&T explains, [w]hen Congress instructs an agency to take reasonable steps to accomplish any goal, it grants the agency considerable discretion,... and courts will grant the agency substantial deference. 279 Although the precise meaning of the phrase all reasonable efforts may depend[] on the circumstances involved, 280 no interpretation requires the Commission to take every conceivable measure to preserve the coverage area and population served of each broadcast licensee See, e.g., Comments of the Association of Public Television Stations, Corporation for Public Broadcasting, and Public Broadcasting Service, GN Docket No , at 9-10 (Jan. 25, 2013); Comments of ABC Television Affiliates Association, CBS Television Network Affiliates Association, FBC Television Affiliates Association, and NBC Television Affiliates, GN Docket No , at 33 (Jan. 25, 2013) ( Broadcast Affiliate Comments ); Comments of the Walt Disney Company, GN Docket No , at (Jan. 25, 2013) ( Disney Comments ); NAB Comments at 24-25; Tribune Comments at 16-17; Harris Corp. Comments at 8 (interpreting the Spectrum Act to require the Commission preserve broadcaster service to all persons that a station currently serves ). 277 Spectrum Act 6403(b) (emphasis added). 278 Id. 6403(b)(2). 279 AT&T Comments at 77 (citing Capital Network Sys., Inc. v. FCC, 28 F.3d 201, 204 (D.C. Cir. 1994)). 280 Id. 281 See generally Kenneth A. Adams, Understanding Best Efforts and Its Variants, The Practical Lawyer (Aug. 2004), (citing Coady Corp. v. Toyota Motor Distrib., 361 F.3d 50, 59 (1st Cir. 2004) ( Best efforts... cannot mean everything possible under the sun. ); Triple-A Baseball Club Assocs. v. Northeastern Baseball, Inc., 832 F.2d 214, 228 (1st Cir. 1987) ( We have 86

95 Instead, the Commission has correctly recognized that the phrase all reasonable efforts... comports with the common meaning of the word reasonable. 282 The D.C. Circuit has interpreted this phrase in a different context to require only feasible actions. 283 And as the Commission notes, the phrase all reasonable efforts does not require the exertion of unreasonable efforts. 284 Further, the reasonableness requirement by its plain terms is a measure of effort i.e. the actions taken to achieve a goal and not of the outcome itself. Congress could have required that the Commission to preserve the coverage area and population served of each broadcast television licensee, but it did not. 285 Even the National Association of Broadcasters recognizes that the phrase all reasonable efforts does not require the Commission to perfectly replicate broadcast contours. 286 Thus, under the plain language of the statute, the Commission only need make reasonable efforts to maintain coverage area and the population served. found no cases, and none have been cited, holding that best efforts means every conceivable effort. ); Bloor v. Falstaff Brewing Corp., 601 F.2d 609, 614 (2d Cir. 1979) ( The requirement that a party use its best efforts necessarily does not prevent the party from giving reasonable consideration to its own interests. )). 282 NPRM 105 (citing Blacks Law Dictionary (6th Ed. 1990) at 1265, which defines reasonable, among other things, as [f]air, proper, just, moderate, suitable under the circumstances. Fit and appropriate to the end in view. ). 283 See Raicovich v. U.S. Postal Service, 675 F.2d 417, 424 (D.C. Cir. 1982); see also Miller v. U.S. Postal Service, 231 Ct. Cl. 804, 810 (Ct. of Cl. 1982). 284 NPRM 105 n Spectrum Act 6403(b)(2). 286 See NAB Comments at 19. Other commenters agree. See, e.g., CTIA Comments at 35 (explaining that requiring precision in this area... would greatly complicate the repacking process and is not required by the Spectrum Act ); CEA Comments at 32 (acknowledging that there may be some situations in which reductions in service areas of more than two percent will occur, and [i]t is reasonable in such cases for the Commission to allow a greater than two percent change in contour or interference level ). However, the Walt Disney Company argues that any reduction in a station s service area constitutes an involuntary relinquishment of spectrum rights in violation of the Spectrum Act. Disney Comments at 34. Not so. Congress never has suggested that the lack of perfect replication constitutes an involuntary relinquishment of spectrum. See Spectrum Act 6403(b)(2). The Commission, moreover, has ample authority to modify licenses to promote the public interest. See Improving Public Safety Communications in the 800 MHz Band, Report and Order, Fifth Report and Order, Fourth Memorandum Opinion and Order, and Order, WT Docket 02-55, FCC , (Aug. 6, 2004) ( 800 MHz Order ). Sections 316, 303, 301, and 154(i) of the Communications Act grant the Commission authority to conduct spectrum-management activities, including license modifications, in furtherance of the public interest. See 47 U.S.C. 316, 303, 301, and 154(i). And the courts have acknowledged and deferred to the Commission s spectrum-management authority and 87

96 2. The Purpose of the Spectrum Act, Including Provisions Addressing Repacking of Broadcasters, is to Reallocate Spectrum for Broadband Deployment. The Commission s repacking methodology should maximize the amount of spectrum that can be repurposed for mobile broadband services. In fact, the animating purpose behind broadcast repacking is to do just that. 287 Numerous compelling public policy reasons justify prioritizing clearance of the band for broadband so long as reasonable efforts are made to preserve broadcast coverage area and population in the repacking effort: First, as Cisco and U.S. Cellular explain, repurposing the broadcast television spectrum for broadband use represents the single greatest opportunity to satisfy burgeoning consumer demand for wireless broadband applications and services. 288 Although the current broadcast television allocation can help satisfy video communications based on a one-to-many model, the allocation cannot satisfy the wealth of data communications based on the many-to-many model of the modern Internet, such as presence information, context-sensitive data, and social networking. The allocation of the nation s spectrum resources must change in response to consumer demand and technology. Second, the continued development and expansion of mobile broadband networks and the adoption of mobile broadband technologies throughout the economy directly contributes to judgment. See 800 MHz Order 64 (citing Teledesic LLC v. FCC, 275 F.3d 75, 84 (D.C. Cir. 2001) ( [W]hen it is fostering innovative methods of exploiting the spectrum, the Commission functions as a policymaker and, inevitably, a seer roles in which it will be accorded the greatest deference by a reviewing court. ) (citation omitted)). Nothing in the Spectrum Act constrains this authority to administer spectrum licenses to promote the public interest. 287 See 158 CONG. REC. H. 914 (2012) (statement of Rep. Upton) ( One of the key elements of [the Spectrum Act] is freeing up an enormous swath of spectrum for use. ), (statement of Rep. Waxman) ( Our bipartisan, bicameral negotiations resulted in legislation that will make new spectrum available for broadband services. ), (statement of Rep. Walden) ( The underlying piece of this legislation frees up spectrum that will generate hundreds of thousands of jobs as 4G is built out. They need spectrum to build out 4G. This provides spectrum. ), available at See also, e.g., AT&T Comments at 74-76; TIA Comments at 6-8; Verizon Comments at See Cisco Comments at 7; U.S. Cellular Comments at 2. 88

97 increases in productivity. 289 Maximizing the amount of reclaimed spectrum will allow for the continued proliferation of advanced mobile networks and devices, resulting in increasing productivity gains and economic growth. Third, as U.S. Cellular details, making available large amounts of spectrum is crucial for the development of greater mobile broadband competition, and with competition comes increased investment and innovation. 290 Fourth, repacking with an eye towards maximizing the amount of spectrum available for flexible use is critical to the success of the incentive auction itself. Under the Spectrum Act, revenues from forward auctions enable the Commission to reimburse repacked broadcasters, pay winning reverse auction bidders, fund a nationwide public safety network, and pay down a portion of the national debt. 291 As AT&T observes, failure to maximize the amount of spectrum recovered for flexible use would dramatically increase the risk of auction failure. 292 Fifth and finally, as the Commission itself notes, addressing America s spectrum challenge is essential to continuing U.S. leadership in technological innovation, growing our economy, and maintaining global competitiveness. 293 Some commenters, however, argue that the Spectrum Act s principal purpose is not to free additional spectrum for broadband use but to preserve the population and service area of existing broadcast channels. 294 Comcast and NBCU, for example, claim the Spectrum Act s 289 Cisco Comments at U.S. Cellular Comments at 3 (citing Fifteenth Report at 9820); Joint Statement on Broadband, GN Docket No , 25 FCC Rcd 3420 (Mar. 16, 2010)). 291 See Cisco Comments at AT&T Comments at NPRM See e.g., Broadcast Affiliate Comments at 19-38; Comcast & NBCU Comments at 11-14; Disney Comments at 34-35; Harris Corp. Comments at 5-9; NAB Comments at 18-31; Comments of Sinclair Broadcast Group, Inc., GN Docket No , at (Jan. 25, 2013) ( Sinclair Comments ); Tribune Comments at

98 directive to make all reasonable efforts to preserve broadcast coverage and population requires the Commission to focus first and foremost on preserving the ability of broadcast stations to continue to serve the needs and interests of their viewers. 295 Comcast and NBCU further claim that the statute does not even permit the Commission to consider how efficiently it is repacking television stations; instead, the Commission must focus solely on preserving the broadcasters coverage areas and populations served. 296 Relying on these (faulty) interpretations, certain commenters seek to impose on the Commission specific requirements for the repacking outcome. For example, NAB proposes that the Commission should preserve service to all of a station s specific viewers, rather than measuring population served with regard to the same total number of viewers. 297 NAB further seeks to have the Commission limit the amount of interference created by an individual channel assignment, considered alone, to a 0.5% reduction in population served, 298 provided further that the Commission caps the total amount of such additional interference at 1% of the total population served. 299 The Commission should reject requests to adopt overly stringent standards for repacking because they are based on fundamentally unsound interpretations of the Spectrum Act. Insisting 295 Comcast & NBCU Comments at Id. ( The Notice... proposes that the all reasonable efforts mandate should be understood to depend on all of the circumstances involved, suggesting that the Commission thinks Congress intended a complex balancing act, with the rights of broadcasters and viewers under the Act in one hand and the efficiency of the repacking process in the other.... Compliance with the Spectrum Act s mandate... requires the Commission to focus first and foremost on preserving the availability of broadcast stations to continue to serve the needs and interests of their viewers. ); see NAB Comments at 19 (interpreting the Spectrum Act to allow the Commission flexibility in repacking only where there are extraordinary circumstances that prevent the Commission from fully preserving a broadcaster s coverage area and population served). 297 NAB Comments at 20; see also NPRM Even absent NAB s further proposed 1% cap, this option for interpreting reasonable efforts is overly restrictive because it measures whether the population served before and after repacking contains the exact same viewers. 299 See NAB Comments at 20-21; see also Tribune Comments at 17; Belo Comments at

99 on the technical identity of viewers or population or all but de minimis variations in coverage area would frustrate the intent of Congress to free additional spectrum for broadband use by greatly complicating, if not altogether thwarting, the repacking. 300 As Verizon explains, although [t]he Spectrum Act... requires the Commission take reasonable efforts to preserve the broadcasters coverage areas and population served, that population served should be based on total over-the-air population, not the same specific viewers, because otherwise, the Commission would be locked into preserving existing geographic markets irrespective of the different radiofrequency environment in which the repacked station would operate. 301 Furthermore, arbitrary caps and interference restrictions not only are inconsistent with the provisions of the Spectrum Act, but also will create severe logistical impediments to the Commission s repacking efforts and prevent new 600 MHz licensees from using the spectrum in an efficient and productive manner. Absent some measure of flexibility to conform broadcast operations to new frequencies and locations, repacking would be impossible and would frustrate the Spectrum Act s purpose of freeing additional spectrum for broadband use. B. The Commission Has a Number of Options to Overcome the Challenges of Repacking Broadcasters. Although repacking the remaining broadcast television licensees poses a number of challenges, 302 there are a variety of measures the Commission can adopt to overcome them. As 300 See CTIA Comments at (explaining that perfectly replicating broadcast contours is not even feasible and would greatly complicate the repacking process). 301 Verizon Comments at 36. CEA agreed, noting that the Spectrum Act s directive to make all reasonable efforts to protect broadcasters existing populations instructs the Commission to strike a balance between replication of coverage area and population served and other considerations and actually prohibits the Commission from seeking to replicate existing populations covered in all instances. CEA Comments at See, e.g., Disney Comments at 38 (noting the time needed for numerous governmental approvals and extensive coordination of construction projects in major urban areas as well as the logistical challenges posed by weather and diffuse facility ownership). 91

100 the Commission explained in the NPRM, repacking involves reorganizing the broadcast television bands so that the television stations that remain on the air... occupy a smaller portion of the UHF band, allowing the Commission to reconfigure a portion of the UHF band into contiguous blocks of spectrum suitable for flexible use. 303 The Commission should address the challenges of broadcast relocation in ways that: (1) maximize broadcaster participation; (2) reduce uncertainty for both reverse and forward auction bidders; (3) expedite the reallocation of spectrum for flexible use; (4) ensure full and timely reimbursement of repacked broadcasters; and (5) reduce opportunities for waste, fraud, and abuse. To address the repacking challenges, the Commission can and should require all broadcasters to provide it with an inventory of their equipment and facilities that will be affected by the repacking process, along with an estimate of the repacking costs. Several commenters proposed that broadcasters be required to conduct an audit of existing equipment and facilities. 304 As CTIA explains, without broadcaster inventory data, the Commission will be unable to determine effectively the transition timelines for repacking nor will it be able to optimize the repacking algorithm to minimize disruptions to incumbent TV stations, and minimize relocation costs. 305 These inventories are important for the Commission to understand the effect and costs of repacking on each broadcaster, because, as Comcast noted, the potential repacking costs for broadcasters can vary substantially depending on the details of the reassignment and the particular facilities of the station involved. 306 By obtaining and verifying these assessments, the Commission would have a more thorough understanding of the nature and scope of the repacking 303 NPRM See CTIA Comments at 35-36; Sprint Comments at CTIA Comments at Comcast & NBCU Comments at

101 costs and logistics and will be able to minimize costs and speed the repacking process. Compiling an inventory of equipment and facilities need not be overly burdensome on broadcasters. As Sprint explains, many broadcasters likely have some of this inventory already completed due to the recent DTV transition, and broadcasters who are contemplating participating in the forward auction have likely begun to compile an inventory, as well as estimates of potential relocation costs. 307 Furthermore, requiring broadcasters to compile an inventory and estimate repacking costs could have other benefits as well, such as reducing the opportunity for waste, fraud, and abuse in the reimbursement process. The Commission should also engage the equipment vendor community in discussions to develop a better understanding of the requirements, costs, and timeframe for completing the broadcaster transition on both a market-by-market and a national basis. 308 T-Mobile supports CTIA s recommendation to gather information from television equipment manufacturers in order to better understand the capabilities of radiofrequency channel modifications to television antennas and transmitters. 309 With this information, along with inventories from the broadcasters, the Commission will be well-positioned to determine the costs associated with the repacking that can be input into its model to ensure the most effective and cost-effective repacking. 310 Sprint similarly observes that reaching out to broadcaster and television equipment vendors will allow the Commission to better understand and scrutinize broadcaster relocation cost estimates. 311 Having this information would also help guide Commission 307 Sprint Comments at See id. 309 CTIA Comments at Id. 311 Sprint Comments at

102 decision-making concerning the timeframes need to complete the repacking process. 312 As with broadcaster inventory information, the lack of input from equipment manufacturers would deprive the Commission of the critical information it needs to set repacking transition timelines, minimize disruptions to broadcasters, and minimize relocation costs to the public. 313 To ensure timely and predictable relocation of broadcasters that relinquish their spectrum, the Commission should adopt firm milestones that a broadcaster must satisfy prior to receiving full payment for relinquishing its spectrum rights or for reimbursement of its relocation costs. 314 As Sprint notes, there are a number of milestone structures and methodologies that the Commission could adopt to achieve this end. 315 Additionally, a framework with intermediate deadlines would be familiar to the broadcast community because infrastructure deployments are typically subject to contractual milestones and performance benchmarks. As Sprint recommends, the Commission should withhold or cancel payment to any broadcaster that fails to meet a specific milestone absent good cause or does not timely transition by a specific date the Commission specifies without a showing of exceptional circumstances beyond the licensee s control. 316 Likewise, full reimbursement should be conditioned on completion of relocation activity (i.e., the final milestone). Accordingly, the Commission should disregard any pleas to make final reimbursement a closing condition of the auction, which could empower broadcasters to unreasonably delay the finality of the auction. 317 These measures will provide forward auction 312 CTIA Comments at Id. 314 See Sprint Comments at Id. at 12 (detailing one milestone option for reverse auction payments and relocation reimbursement costs that would make 50 percent of payments to reverse auction participants at the conclusion of the auction; 25 percent upon the execution of contracts between suppliers and broadcasters in cases where a broadcaster is relocating, channel sharing, or being repacked; and the final 25 percent when the spectrum is made available for mobile broadband use). 316 See id. 317 See Sinclair Comments at

103 winners with reasonable confidence that spectrum won at auction will be available by a date certain. 318 Without the assurance of reasonable benchmarks for broadcast clearing, fewer parties may participate in the forward auction and those that do participate may bid less aggressively, which would reduce auction revenues and fail to increase competition in the mobile broadband market. The Commission should also establish reimbursement payments based on the broadcasters inventory and cost estimates that are informed by discussions with vendors to prevent the escalation of reallocation costs beyond the statutorily authorized amount of money available for relocation. 319 Several commenters proposed a sensible two-step reimbursement process in which the broadcaster would receive an initial upfront payment followed by a true up, a process the Commission has successfully permitted before and one which could prove helpful to ensure all expenses are fully reimbursed while safeguarding against fraud and abuse. 320 As explained by the Tribune Company, [t]his two-step approach would ensure that broadcasters could relocate or modify facilities as quickly as possible without concern for whether they can secure the necessary capital to effectuate the repack. 321 Public broadcasting commenters note that this approach will ensure that broadcasters encountering unexpected challenges, such as severe weather events and other disruptions that can significantly increase costs and make it more difficult to accurately estimate a station's relocation costs," are able to be made whole for 318 See Sprint Comments at See id. at See Comments of Association of Public Television Stations, Corporation for Public Broadcasting, and Public Broadcasting Service, GN Docket No , at (Jan. 25, 2013) ( Public Broadcaster Comments ); Comcast & NBCU Comments at 24; NAB Comments at 53-54; Tribune Comments at Tribune Comments at

104 their actual relocation costs. 322 Comcast agrees, noting that adopting this two-step approach is both within the Commission s authority and area of expertise as the Commission has used such an approach with respect to reimbursement credits issued to Nextel in the 800 MHz proceeding. 323 And NAB details additional advantages of the two-step reimbursement approach, including treating all broadcasters equally and ease[ing] the [upfront] burden of capital financing for repacking expenses. 324 Finally, requiring all broadcasters to document actual expenses would not only reduce the likelihood of waste, fraud, and abuse, but also increase the odds of identifying and sanctioning improper expenses. 325 While a two-step cost recovery holds great promise for timely and effective spectrum clearing, the Commission would still need to exercise close oversight over the process to guard against strategic delays, gold plating, and other behaviors that might escalate costs or extend the relocation process. Derivations from presumptively reasonable amounts and transition schedules should be the exception, not the rule. Protecting only those facilities licensed, or fully eligible for licensing, as of February 22, 2012, provides the certainty forward auction bidders need and reverse auction participants have a right to expect. 326 The Spectrum Act identifies February 22, 2012 as the date before which the Commission must make all reasonable efforts to preserve the coverage area and population of each broadcast licensee. 327 Tying the application cut-off date to the statutorily established cutoff date for contour protection avoids the inequities of having those broadcasters with licenses granted or pending prior to February 22, 2012 receive more protection than those that come later 322 Public Broadcaster Comments at Comcast & NBCU Comments at NAB Comments at Id. 326 TIA Comments at 8; T-Mobile Comments at 37; NPRM Spectrum Act 6403(b)(2). 96

105 in the licensing process. While some broadcasters state that the February 22, 2012 date proposed by the Commission is problematic or unfair, nothing suggests the date is arbitrarily over- or under-inclusive, and the Commission would likely face similar complaints regardless of the date selected. 328 The alternative of a less definitive or more flexible cut-off date is far worse. Failing to establish an unambiguous cut-off date for broadcast participation risks delaying or disrupting the auction by making a moving target out of the spectrum that is its subject. Finally, as advocated by the Tribune Company, the Commission should support measures to encourage the Internal Revenue Service ( IRS ) to ensure that reimbursed reverse auction relocation expenses and relocation proceeds are treated equitably from a tax perspective. 329 The Commission should not delay resolution of this proceeding while waiting for an IRS ruling, but it should work with the IRS to allow any gains from a broadcaster s participation in the reverse auction to be treated as an involuntary conversion, with the proceeds to be deferred for income tax purposes. As the Tribune Company explains, if the IRS were to consider the auction an involuntary disposition of the broadcasters caused by government action, broadcasters may defer any gains from their participation in the reverse auction so long as those gains are reinvested in similar property. Furthermore, the Commission should attempt to work with the IRS to provide clarification that any gains attributable to reimbursement for repacking may be deferred. 330 By working with the IRS to ensure favorable tax treatment for participating broadcasters, the Commission can encourage greater broadcaster participation, helping it achieve 328 See e.g., id.; Network Affiliate Comments at 21-24; Comments of Bahakel Communications, Ltd., GN Docket No , at 1-3 (Jan. 25, 2013); Comcast & NBCU Comments at 14-16; Comments of SATV10 LLC, GN Docket No , at 3-4 (Jan. 25, 2013); Comments of Univision Communications Inc., GN Docket No , at 8-13 (Jan. 25, 2013). 329 See Tribune Comments at See id. at

106 its goal of maximizing the amount of spectrum reclaimed for flexible use. 331 Moreover, by eliminating the need for broadcasters to factor their tax burdens into their reverse auction bids, the Commission can promote a more efficient auction where reverse auction bids more accurately represent demand. C. Secondary Means Secondary. Secondary licenses must not impede the clearing and repacking process. Secondary licenses are not secondary to only some primary licenses, but secondary to all primary licenses. By longstanding Commission rule and practice, secondary licenses receive no protection against interference from primary users and must resolve any interference caused to new, existing, or modified primary users, including going off the air if necessary. 332 Therefore, secondary licenses must give way to primary licensees old and new. 333 As secondary licenses, lower power television stations are not entitled to receive the same protection as primary licensees during or after the repacking process. 334 Nevertheless, a handful of commenters claim that secondary stations are secondary only to full-power television licenses, rather than secondary to all licenses, including newly awarded mobile broadband licenses Id. at NPRM 74 (citing 47 C.F.R and ; Digital Low Power Television, Television Translator, and Television Booster Stations and Digital Class A Television Stations, Report and Order, 19 FCC Rcd 19331, (2004) ( DLPTV Report & Order ). 333 DLPTV Report & Order, Appendix B See e.g., CTIA Comments at 37; PISC Comments at 6, 52-55; TIA Comments at 7-8; Comments of the Wireless Internet Service Providers Association, GN Docket No , at (Jan. 25, 2013); TIA Comments at 8 (citing 47 U.S.C. 303, 306) ( noting the Commission s longstanding statutory authority to distinguish between primary and secondary [spectrum] uses and asking the Commission to rescind the licenses of non-class A low-power TV stations or other secondary users where doing so facilitates efficient repacking in order to maximize the amount of reclaimed broadcast spectrum. ). 335 See e.g., Comments of the Advanced Television Broadcasting Alliance, GN Docket No , at 2, 5 (Jan. 25, 2013); Comments of DTVAmerica Corporation, GN Docket , at 2 (Jan. 25, 2013); Comments of Mako Communications, LLC, GN Docket , at 3 (Jan. 25, 2013); Comments of MSGPR Ltd. Co., GN Docket No , at 2 (Jan. 25, 2013). 98

107 These claims are baseless. The Commission s rules make no such distinction. 336 On the contrary, the Commission made clear more than three decades ago that secondary, low power television stations may not cause interference to, and must accept interference from, full-service television stations, certain land mobile radio operations and other primary services. 337 Low power television stations are secondary stations with secondary status to all primary services in the spectrum. D. Authorizing Unlicensed Use of 600 MHz Bands Prior to Commercial Deployment Will Decrease Auction Revenues and Delay Broadband Deployment. Allowing whites space and other unlicensed devices to operate on newly cleared spectrum before wireless companies have had the full opportunity to deploy services will increase uncertainty surrounding the value of spectrum to be auctioned, decrease auction revenues, and complicate as well as delay broadband deployment. The Commission discussed whether, following the build-out term, [it] should permit third parties to make use of unused spectrum on a localized basis until a licensee deploys service in those areas a use-it-or-shareit build-out requirement. 338 Tellingly, the Commission did not seek comment regarding such a use-it-or-share-it approach to spectrum during the licensee s build-out term. This refusal likely shows that the Commission recognized the many problems a use-it-or-share-it regime would create for carriers building out their spectrum. Rather, the Commission only contemplate[d] applying use it or share it after the licensee s build-out term [is] concluded See 47 C.F.R ; see also DLPTV Report & Order, 19 FCC Rcd at (noting that [s]tations in the low power television service are authorized with secondary frequency use status. ). 337 Id. (citing 47 C.F.R , , ) (emphasis added). 338 NPRM Id. 405 n.625 (citing Letter from Michael Calabrese, PISC, to Marlene H. Dortch, Secretary, FCC, ET Docket No , WT Dockets No , 12-69, 10-4 at 3 (filed August 20, 2012)). 99

108 Nevertheless, a small group of commenters ask the Commission to authorize unlicensed use of the newly purchased and cleared spectrum prior to the end of the build-out term. 340 Google and Microsoft, for example, ask for the Commission to enable unlicensed spectrum operations in areas where a licensee has yet to deploy its network or has ceased operations. 341 These proposals are misguided. For one, the white space-model is ill-suited for commercial broadband deployment. Cellular terrestrial systems use many more base stations than single-transmitter broadcast systems. During the construction and deployment of services, mobile companies are required to test many different geographic areas, whereas broadcast stations remain stationary. As CTIA explains, [r]equiring the licensee to share its spectrum with other uses while in the process of expanding into new geographic areas would undermine or delay the provision of service in these areas. 342 In particular, operation of unlicensed devices would interfere with a licensee s ability to test and build out its network, two integral steps in commercial deployment. 343 Even when the geographic area is not technically built out, mobile companies need that spectrum to be available for testing throughout the area. Furthermore, identifying and using white spaces in a nascent terrestrial wireless band, such as the 600 MHz band, would impose new notification and clearance challenges on auction winners. The uncertainty surrounding how any notification system would work and the significant and complicated expenses associated with such notification will decrease the value of 340 See Google and Microsoft Comments at 44; Comments of the White Space Database Administrators, GN Docket No , at 3 (Jan. 25, 2013); Comments of WhiteSpace Alliance, GN Docket No , at (Jan. 25, 2013); Comments of Spectrum Bridge, Inc., GN Docket No , at 2 (Jan. 25, 2013). 341 Google and Microsoft Comments at CTIA Comments at Id. 100

109 the spectrum. Moreover, there is further risk that once operating on the licensed spectrum, unlicensed devices will not leave the spectrum, whether because the devices can potentially operate under the radar for a period of time, the device owners never received notification, or companies commenced legal and lobbying battles for a continued right to use the spectrum. In any event, as the National Telecommunications Cooperative Association explains, use-it-orshare-it proposals are of limited utility. 344 For these reasons, CTIA properly concluded that the substantial uncertainty... as to whether the band would be available when needed would result in delayed broadband deployment to consumers. 345 VI. CONCLUSION The incentive auction of television broadcast spectrum offers the potential for many improvements to the wireless market, including enhanced competition and more efficient deployment of services to consumers. However, it also poses numerous challenges from the myriad logistical and technical issues posed by any given plan to the competing interests of the various market participants, including wireless providers, broadcasters, and equipment manufacturers. In striking a balance, the Commission should adopt those options that have emerged from the comments as best achieving the Commission s goals with the least trade-offs. Specifically: The band plan: A band plan that incorporates at least 35x35 MHz of paired spectrum through uplink and downlink bands in paired 5x5 MHz blocks above Channel 37 provides the most advantages with the fewest and most manageable shortcomings. This plan maximizes the availability of paired spectrum, draws on currently-existing technology while providing room for 344 Nat l Telecom. Coop. Ass n Comments at CTIA Comments at

110 innovation, limits harmful interference, and promotes competition. Neither the Commission s lead plan nor alternative plans advanced by various commenters can accommodate as many of these goals with as few countervailing issues. The forward auction: forward and reverse auctions. A successful auction requires robust participation in both the The market power of the nation s two largest carriers could compromise broad participation in the forward auction unless the Commission institutes safeguards to prevent these two carriers from further spectrum aggregation. The Commission should adopt forward auction policies that enhance competition and bidder participation, including a spectrum cap on the high-value low-frequency beachfront spectrum below 1 GHz. The reverse auction: A simplified reverse auction that sets forth a clear process for participation will provide the transparency necessary to mitigate broadcaster concerns about participation and therefore result in an auction that maximizes the amount of available spectrum for broadband use. The more certainty the Commission can provide the reverse auction process (and the more quickly the Commission can provide it), the better. The repacking: The Commission can undertake reasonable efforts in the relocation of broadcasters following the reverse auction that will sufficiently protect those broadcasters while also serving the Spectrum Act s primary goal of reallocating spectrum for broadband deployment. The Commission need not and cannot achieve a perfect outcome in which each broadcaster has the same coverage and population it had prior to the auction. However, the comments have identified numerous safeguards the Commission can apply to ensure that broadcasters are made whole for their relinquishment of spectrum and reasonable relocation expenses. 102

111 The Commission should move swiftly to finalize its band plan, auction rules, and repacking methods in anticipation of delivering much-needed low-frequency spectrum to the wireless market in Respectfully submitted, /s/ Thomas Sugrue Ari Fitzgerald Trey Hanbury Phillip Berenbroick AJ Burton Hogan Lovells US LLP 555 Thirteenth Street, NW Washington, DC (202) Attorneys for T-Mobile USA, Inc. Thomas Sugrue Kathleen O Brien Ham Steve Sharkey Christopher Wieczorek Indra Chalk Joshua Roland T-Mobile USA, Inc. 601 Pennsylvania Avenue, NW Washington, DC (202) March 12,

112 Exhibit A

113 Roberson and Associates, LLC Technology and Management Consultants Analysis of the 35x35 MHz Band Plan Proposal for 600 MHz Spectrum Roberson and Associates, LLC Contributors: S. Borkar D. Roberson K. Zdunek March 11, 2013 FINAL

114 Analysis of the 35x35 MHz Band Plan Proposal for 600 MHz Spectrum Executive Summary This whitepaper presents an analysis of the technical feasibility of the 35x35 MHz band plan proposed by T-Mobile, USA for the MHz band. The plan occupies a total of 84 MHz of spectrum beginning at 614 MHz, the upper boundary of channel 37, and continuing to 698 MHz, and includes a 4 MHz guard immediately adjacent to channel 37 and a 10 MHz duplex gap between downlink and uplink. The analysis was conducted by developing a set of core evaluation criteria based on principles put forth by the FCC in its October 2012 NPRM and by industry in its January 2013 letter to the FCC on Expanding the Economic and Innovation Opportunities of Spectrum through Incentive Auctions. The core evaluation criteria can be summarized as follows: Utility and Quantity: o Maximize the traffic capacity available to users, and maximize the opportunity for wireless service providers. o Provide for equal amounts of uplink and downlink spectrum. Technical Feasibility and Cost Competitiveness o Utilize current technology, or employ technology advances that can expected to be achieved in o Achieve technical performance that does not significantly impact traffic capacity or user operation, and can be provided at competitive cost. Do No Harm o Avoid, eliminate or render inconsequential, interference with incumbent users such as cellular or broadcast TV. Interchangeable Spectrum Blocks. Harmonized and Interoperable Band Plan. Where technical performance issues are identified, the analysis seeks to identify feasible solutions that mitigate or eliminate any adverse effects. The results of the analysis reveal that the 35x35 MHz plan shares several technical issues in common with other proposed plans, none of which prevents successful operation or deployment. These issues include: Interference due to TV Operation in the uplink segment of the band. The size of the Duplex Gap and the types of operation allowed in the gap. The guard band size required to protect the cellular operations from broadcast TV operations. The need to address interference between cellular and TV operations in geographically adjoining and national border regions. Maintaining equal relative value of the spectrum blocks. FINAL 2 3/11/2013

115 Importantly, similar to other band proposals, the 35x35 MHz plan removes technical issues inherent to the FCC proposed band plan, in particular, spectral inefficiency and inter-modulation distortion potential resulting from operation of broadcast TV in the duplex gap. The analysis has also identified performance characteristics unique to the 35x35 MHz plan compared to other plans. The most important of these characteristics is the achievement of a nominal 40% increase in wireless broadband capacity compared to a 25x25 MHz band plan. Other characteristics associated with the 35x35 MHz band plan involve issues of handset antenna performance, duplex filter feasibility, and the potential for harmonic interference with a portion of the existing PCS bands. Our analysis shows that current technology or technology improvements expected by can resolve, or substantially mitigate, any performance issues that result from the 35x35 MHz configuration T-Mobile has proposed. The challenges associated with the 35x35 MHz configuration, their related performance issues, and resolution approaches are summarized below: Technical Question Single device antenna covering wide bandwidth Performance Issue(s) or Concern(s) Antenna Length Increase causes db detuning Antenna efficiency causing degraded detuning of -0.6 db Larger antenna size compatible with handset Resolution Approach(es) Optimize antenna for 600 MHz uplink Increase base power Advanced antenna design Result Capacity significantly exceeds other plans Negligible throughput decrease Minor antenna structure size increase-compatible with (4-5 inch smartphones) Advanced antenna designs resolve issues Duplex Filter Sufficient Transmit /Receive Isolation in Handset Harmonic Interference 3.5% Interference with PCS band receiver (3 rd harmonic) Minimal Interference with BRS band (4 th harmonic) Overlapping duplex filter structure Advanced duplexer materials Frequency coordination Improved RF harmonic filter Acceptable performance achieved Slight cost increase mitigated by unit volumes Advanced duplexer approaches resolve issues No degradation to PCS or BRS band device or user Again, taking these challenges into account, the 35x35 MHz band plan still achieves a nominal 40% capacity increase compared to a 25x25 MHz band plan and an effective spectrum utilization efficiency of 88% (70 MHz out of 80 MHz including the duplex gap). Positive attributes of the 35x35 MHz plan shares with other band plans include: FINAL 3 3/11/2013

116 Creating 5 MHz paired interchangeable FDD blocks Allowing channel 37 ( MHz) to continue to be used for medical and other unlicensed applications and Providing a total of 10 MHz guard band for broadcast TV channels below 37 (608 MHz) Other characteristics that the 35x35 MHz band plan shares in common with other proposed 600 MHz band plans, including 25x25 MHz and 30x30 MHz band plans, include: The downlink band ( MHz) is intended to be reserved in all geographic areas. However in the uplink band ( MHz), TV stations may still be allowed to operate in some geographic regions, with the guard band used to protect cellular operations in these regions Impact of less spectrum freed up in uplink region in certain geographic and border areas: o Mitigation through geographic isolation, co-siting, antenna orientation coordination Impact of 600 MHz operation on size of the device antenna: o 10% larger than 700 MHz Operation Concurrent support for multiple bands, e.g., 700 MHz, AWS, PCS The spectrum below channel 37 (608 MHz) can be used for TV transmission, Supplementary Downlink (SDL), or as unpaired LTE using Time Division Duplex (TDD) technology Duplex gap can be used in a secondary fashion for unlicensed and other non-interfering applications The analysis shows that none of the challenges in the above characteristics impose technical requirements which cannot be met using current technology. In summary, the technical performance requirements of the 35x35 MHz plan can be met by current and reasonably anticipated emerging technologies. The 35x35 MHz band plan provides the largest amount of broadband wireless capacity compared to the other band plans that have been proposed. Instead of being limited by current traditional technologies, it leverages anticipated technical advances and offers a longer range, cost effective solution for the 2014/2015 timeframe. FINAL 4 3/11/2013

117 Table of Contents 1.0 Introduction MHz Incentive Auction Purpose and Scope of the Whitepaper Organization of the Whitepaper Analysis Approach and Band Plan Evaluation Criteria FCC Criteria Industry Criteria Core Criteria Description of the 35 x 35 MHz Band Plan Band Plan Reference Terms x35 MHz Band Plan Analysis of the Proposed 35x 35 MHz Band Plan Strengths of the Proposed Band Plan : Technical Issues Unique to the 35x35 MHz Band Plan and Their Resolution : Antenna Design : Duplex Filter Design : Harmonic Interference Background Analysis of Harmonic Interference : Issues Common to All Major Proposed Band Plans : Duplex Gap : Interference due to Broadcast TV Operations : Geographically Overlapped Region Adjacent Channel Interference : Geographically Overlapped Region Guard Band : Geographically Adjoining and Border Regions Co-Channel Interference : Geographically Adjoining and Boundary Regions Adjacent Channel Interference : Interference Impact Summary Conclusions References...46 Appendix A: Other Band Plan Proposals...49 A.1 25x25 MHz Plan...49 FINAL 5 3/11/2013

118 A.2 30x30 MHz Plan...50 Appendix B: Company Profile...51 List of Figures Figure 1: Cellular Band Plan Parameters...12 Figure 2: Proposed 35x35 MHz Band Plan...12 Figure 3 Reference Diagram for Device Antenna...15 Figure 4: Duplex Filter...19 Figure 5: Filter Segmentation Illustration...20 Figure 6: Two 25 MHz Overlapping Filters...20 Figure 7: Illustrative Power Levels of Transmitter Harmonics...22 Figure 8: Harmonic Interference into the PCS Band...23 Figure 9: Third Harmonic Interference Due to LTE Block Characteristics...24 Figure 10: TV Co-Channel and Adjacent Channel Interference...28 Figure 11: Targets of Potential Interference...30 Figure 12: Guard Band for Minimizing TV Interference...31 Figure 13: 10 MHz Guard Band between Band 12 and TV Channel 49 [ALU]...32 Figure 14: Cell Edge Area...33 Figure 15: Moderate Power TV Operation with Guard Bands...36 Figure 16: High Power TV Operation with Guard Bands...36 Figure 17: Illustration for TV Operations in 25x25 MHz Uplink...37 Figure 18: Illustration for TV Operations in 35x35 MHz Uplink...37 Figure 19: Geographically Adjoining Areas...38 Figure 20: Co-Channel Spectrum Overlap...39 Figure 21: Capacity Loss due to Co-Channel Interference...40 Figure 22: Capacity Loss as a Function of Distance between TV Station and Base Station...41 Figure A.1: A Representative 25x25 MHz Band Plan...49 Figure A.2: A Representative 30x30 MHz Band Plan...50 FINAL 6 3/11/2013

119 List of Tables Table 1: Antenna Length...16 Table 2: Antenna Tuning Options...17 Table 3: Harmonics for the MHz Uplink Transmission...22 Table 4: TV Interference Scenarios*...29 Table 5: Interference Victims...30 Table 6: Cellular Capacity Loss near Cell Edge...33 Table 7: Additional Mitigation Approaches...34 Table 8: Capacity Reduction for TV Stations in Uplink Region...37 Table 9: TV Operations in Geographically Overlapped Regions...42 Table 10: TV Operations in Geographically Adjoining and Border Area Regions...43 FINAL 7 3/11/2013

120 1.0 Introduction In response to the FCC NPRM document released in October 2012 [FCC], several comments and Band Plan proposals have been submitted. T-Mobile proposed a 35x35 MHz Band Plan as an effective solution for maximizing the benefit to the end customers and providing a competitive landscape to the operators [TMOB1]. This study provides a technical analysis of this Band Plan proposal MHz Incentive Auction The FCC initiated action via an NPRM describing its intent to conduct an incentive auction to reclaim broadcast TV spectrum in the 600 MHz band and re-allocate that spectrum to commercial broadband wireless service. The intent is for the DTV operators to vacate, using a reverse auction approach, a range of spectrum starting downward from 698 MHz, and for the cellular operators to introduce LTE based services (forward auction) in the vacated spectrum. Additional background on the proposed action is summarized as follows: The FCC included a proposed band plan in its NPRM FCC enumerated basic principles in the NPRM: utility, certainty, interchangeability, quantity, and interoperability and assignments of blocks in 5 MHz sizes Several organizations have submitted responses to the FCC NPRM, critiquing the FCC Band Plan, and offering criteria/principles for developing a high value approach Organizations have also submitted their own band plans intended to overcome technical issues including co-existence with broadcast TV and other cellular systems. T-Mobile submitted a response to the FCC proposing a 35x35 MHz band plan 1.2 Purpose and Scope of the Whitepaper The intent of this whitepaper is to evaluate the technical merits of the 35x35 MHz band plan proposed by T-Mobile. [TMOB1]. The T-Mobile proposal uses the core principles arrived at as a consensus between NAB, cellular operators, and product suppliers [ATT2] and suggests that the approach addresses technical issues including various types of interference, spectrum usage efficiency, and complexity of antenna and filter designs. 1.3 Organization of the Whitepaper The framework including requirements for an effective Band Plan meeting the FCC criteria and the industry consensus is provided in Section 2. The 35x35 MHz Band Plan proposed by T-Mobile is reproduced in Section 3 for the sake of completeness. Section 4 provides a detailed analysis of the 35x35 MHz band plan in terms of its strengths along with the key issues and mitigation approaches. The issues are segmented into two categories those which are common and to be addressed by all other similar Band Plans for the 600 MHz spectrum, and the others which are primarily related to the proposed 35x35 MHz Band Plan. Section 5 provides concluding remarks summarizing the evaluation of the 35x35 MHz Band Plan and how it provides for the criteria set in Section 2. References cited in the study follow. Appendix A reproduces the representative 25x25 MHz and 30x30 MHz Band Plans that have been proposed during the NPRM responses to FCC. The Roberson and Associates company profile is provided in Appendix B. FINAL 8 3/11/2013

121 2.0 Analysis Approach and Band Plan Evaluation Criteria The overall approach to the analysis of the 35x35 MHz band plan is straightforward: identify the relevant evaluation criteria, and then assess the performance of the plan compared to those criteria. An effective and practical band plan would meet the evaluation and performance criteria thus established. The criteria used were derived from two external sources: the October 2012 FCC Incentive Auction NPRM [FCC], and the January 2013 industry Letter to the FCC [ATT2]. The criteria contained in these sources were then consolidated into a set of core criteria, to which additional items addressing technical feasibility and cost/competitiveness issues were added. The FCC criteria, industry criteria, and core criteria are described in detail below. Together, these constitute the framework or reference for evaluation of the 35x35 MHz band plan. 2.1 FCC Criteria The FCC has provided a set of fundamental principles in its NPRM which forms the foundation for developing and evaluating band plans. A sound interpretation of these items is provided in [ATT1], and is summarized and restated below along with some additional clarification: FCC Criteria Utility: maximize the capacity and utility (value) of the available spectrum. This implies that the bandwidth allocated to the LTE cellular operations and the spectrum usage efficiency need to be maximized with appropriate trade-offs in implementation feasibility and complexity. It is also necessary that the interference impact to incumbent users be minimized. Certainty: provide potential spectrum license bidders with certainty as to the characteristics (performance potential) of the spectrum on which they are bidding. Interchangeability: develop spectrum blocks for operators that have similar utility (value) so that they are interchangeable from the point of view of the provider. Quantity: maximize the amount of spectrum bandwidth made available by the auction. For example, this would imply the largest potential operating bandwidth and least amount of overhead (due to guard bands or duplex gaps, for example) Interoperability: support allocation of spectrum in a manner that permits the development of band classes that support equipment and devices that can be used nationally as broadly as possible. 2.2 Industry Criteria Several of the key stakeholders recently sent a consensus document stating its core principles regarding the incentive auction spectrum to the FCC [ATT2]. These principles were based primarily on an analysis of the band plan proposed by the FCC in its NPRM. They identified the technology FINAL 9 3/11/2013

122 implications of the FCC plan and suggested band plan principles that avoid certain technical challenges. The industry criteria are listed below: Industry Criteria Adopt a contiguous down from Channel 51 approach with uplink at the top of the band and continuing downward from 698 MHz Maximize the amount of paired spectrum above TV 37 (rely on supplemental downlink configurations where spectrum is cleared but pairing options are not viable). These unpaired supplemental downlink blocks can be used in concert with currently defined uplink bands in the 700 MHz and PCS bands etc. Rely upon 5 MHz spectrum blocks as building blocks for the band plan Incorporate a duplex gap or spacing between uplink (mobile transmit) and downlink (base transmit) of a minimum of 10 MHz, but no larger than technically necessary. Uplink and downlink filter feasibility dictates a small size limitation Large duplex separation has adverse impact on antenna design [RIM] and user device size [ATT1] Avoid broadcast television stations in the duplex gap. This avoids various potential interference scenarios. Use of guard bands increases the duplex gap and reduces the amount of spectrum available for wireless operation. Preclude any operations in the duplex gap or guard bands that would result in harmful interference to adjacent licensed services Provide guard bands that are consistent with the principle no larger than is technically reasonable to guard against harmful interference between adjacent operations. Provide a guard band between a high power broadcaster and mobile downlinks that is sufficient to protect the wireless service from interference, which will likely be larger than the 6 MHz proposed by the FCC Permit existing operations in TV 37 to remain. This includes the Wireless Medical Telemetry System (WMTS) and radio astronomy applications Facilitate international harmonization, prioritizing harmonization across North America, and move forward expeditiously to coordinate with Canada and Mexico for new broadcast assignments. Harmonization would provide enormous benefits to consumers in terms of reducing costs and increasing the availability of services [RIM]. The primary implication of this requirement is on the duplex gap to be in the MHz range as the accepted norm and on mitigation of co-channel and adjacent channel interference for the cellular operation. 2.3 Core Criteria The following core criteria, derived from and in addition to above sets of criteria, constitute the framework for evaluating the 35x35 MHz band plan. Utility and Quantity Maximize spectrum utilization efficiency to maximize capacity and benefit for users Provide largest amount of spectrum capacity for broadband wireless service Provide for supplementary downlink (SDL) spectrum below channel 37 starting downward at 608 MHz as appropriate. SDL can also be adjacent to the core downlink spectrum [MOTO]. SDL can be used in concert with either the 600 MHz core operation or to supplement the downlink for cellular operation in other bands FINAL 10 3/11/2013

123 Technical Feasibility at Competitive Cost Technical performance, especially relating smartphones and tablets for the major hardware components, including filters, duplexers, tuners, and antenna systems, should be consistent with availability [QUAL1], and be cost competitive Optimize uplink performance (efficiency) over downlink, if necessary, since uplink power is limited and the base stations have more range and capabilities for higher power transmission Single device antenna structure for operating bands < 1 GHz Minimum duplex gap resulting in a much smaller antenna bandwidth and, therefore, a simplified antenna design [MOTO]. Do No Harm Avoid, eliminate or render inconsequential, any adverse effect of harmonics due the 600 MHz LTE uplink transmitter limitations. Avoid, or render inconsequential, broadcast TV coexistence (interference) issues. Interchangeable Spectrum Blocks Keep the downlink spectrum band consistent nationwide and a fixed allocation for the downlink that is cleared of all broadcast stations nationwide [FCC], [CEA], [MOTO]. This minimizes the space and power requirement for downlink filters in mobile devices and helps ensure interoperability across the band [RIM]. Flexible enough to accommodate varying amounts of spectrum relinquished from the incentive auction process in different locations. Harmonized /Interoperable Band Plan Approach The band plan needs to be consistent with the use of advanced LTE techniques including carrier aggregation and Multiple Input Multiple Output (MIMO) and other Advanced LTE Release 10 features. 3.0 Description of the 35 x 35 MHz Band Plan After introducing the basic terminology relevant to a band plan analysis, this section describes the 35x35 MHz plan proposed by T-Mobile USA. The plan provides for commercial broadband wireless operations in the 600 MHz band, using spectrum made available via an incentive auction. The band plan was devised with the intent of satisfying the FCC and industry criteria described in Section 2, in particular, maximizing the available full duplex bandwidth to users and operators in the spectrum region above Channel Band Plan Reference Terms In order to provide a reference for the description and analysis of the band plan, the key terms used in this whitepaper are illustrated in Figure 1. These terms (technical parameters) include: center frequency of operation, pass band width, duplex gap, and total operating bandwidth. Downlink and FINAL 11 3/11/2013

124 uplink refer to the spectrum used for downlink (base station to user device), and uplink (user device to base station) communications, respectively. Figure 1: Cellular Band Plan Parameters Spectrum usage efficiency is defined as (DL + UL)/ Operating Spectrum x35 MHz Band Plan The specific 35x35 MHz band plan proposed by T-Mobile [TMOB1] and the subject of further analysis is illustrated in Figure 2. The band has its 35 MHz uplink defined from 698 MHz downward, followed by a duplex gap of 10 MHz, a 35 MHz downlink, and a 4 MHz guard band, ending at 614 MHz. Figure 2: Proposed 35x35 MHz Band Plan The 4 MHz guard band immediately above channel 37 enables an overall guard band of 10 MHz for any broadcast TV or other operations below 608 MHz. Since Channel 37 is being used for radio astronomy and intra hospital radio communications, these uses are not expected to cause measurable interference into the LTE system in the spectrum above 618 MHz. FINAL 12 3/11/2013

125 The region below channel 37 (608 MHz) may be used for Supplementary Downlink (SDL) broadband wireless operations, TV stations, or LTE Time Division Duplex (TDD) operations. SDL is particularly suited for carrier aggregation or as supplemental downlink. Section 4 describes the analysis of this 35x35 MHz band plan, along with the technical issues that need to be addressed in order to make this plan feasible. The issues are addressed in two categories, 1) those issues unique to this band plan due to its wider bandwidth, and 2) those issues which are common to all the band plans. 4.0 Analysis of the Proposed 35x 35 MHz Band Plan Using the core criteria indicated in Section 2, an evaluation of the 35x35 MHz band plan was performed. Section 4.1 identifies the positive attributes of the plan and highlights the advantages, primarily related to the availability of a greater amount of spectrum. The requirement for a wide band plan in any spectrum region including 600 MHz spectrum may create issues primarily driven by current technology limitations and the potential for interference into other operating cellular spectrum. Technical issues specific to the 35x35 MHz band plan (compared to other band plans) are identified in Section 4.2, along with solutions to mitigate performance issues where they exist. This facilitates introduction of the band plan into the market place. Additionally, issues in common with other band plans that introduce cellular LTE operation in the 600 MHz spectrum have been identified. 1 While most of these common issues have already been discussed in previous submissions to the FCC [ATT1], [ALU], [MOTO], [QUAL], [TMOB1], [VERI], the key issues, the estimates of their impact, and approaches to their resolution are summarized in Section Strengths of the Proposed Band Plan The strengths of the 35x35 MHz band plan can be demonstrated by comparing its attributes to the core evaluation criteria. Utility and Quantity The major value of the 35x35 MHz band plan is its provision for the largest uplink and downlink pass bands, compared to other plans as described in Appendix A. This maximizes the capacity, spectrum bandwidth, and utility of the available spectrum. Simply stated, the band plan creates benefit for the largest number of users. Economies of scale due to high device volumes, and the motivation to develop technology improvements to optimize performance will provide low cost solutions for customers along with high throughput and capacity. The band plan also creates the greatest business opportunity for carriers, operators, and equipment suppliers and fosters a competitive environment. The auction of a wider spectrum range is likely to allow winning bids by a larger number of operators. Another way to capture the value of the 35x35 MHz band plan is to carry out a comparison with the25x25 MHz band plan using key attributes like spectrum availability to users. In order to carry out a fair comparison, the 25x25 MHz band plan is also assumed to have a 10 MHz duplex gap 1 Other band plans, representative of others that have been proposed for the 600 MHz band are described in Appendix A. FINAL 13 3/11/2013

126 instead of 14 MHz [ATT1]. Also, the overhead of the guard Band below the band plan is not included in the calculations since the vacant spectrum region above channel 37 may be used for supplementary Downlink (SDL) or other low power unlicensed applications. The 35x35 MHz band plan provides a nominal 40% increase in useable bandwidth (pass band) as a standalone band as compared to the 25x25 MHz plan and hence makes available a significantly larger amount of data capacity to end users. It utilizes an operating bandwidth of 80 MHz (including the Duplex Gap (DG) of 10 MHz) above channel 37 which is close to the upper limit of 84 MHz. The spectrum usage efficiency correspondingly improves from 83% to 88% as compared to the 25x25 MHz band plan since the Duplex Gap becomes a smaller percentage of the operating spectrum. The characterization of the 35x35 MHz band plan is predicated on the assumption that technology will support the availability of the active pass band spectrum defined in the band plan. This is addressed in Section 4.2 where it is shown that solutions exist for the identified antenna, filter, and harmonic interference issues which can impact the effective pass band spectrum available to the users. The tradeoff in enabling a wider spectrum is expected to require marginally additional hardware, space, and cost. 4.2: Technical Issues Unique to the 35x35 MHz Band Plan and Their Resolution Devising a new band plan is a challenging task that requires analysis and tradeoffs in the design of the transceiver components and technical performance necessary to operate successfully in the band. The key technical issues which have been determined to be in common for all the major band plan proposals for the 600 MHz spectrum, including for 35x35 MHz band plan, are described in the subsequent Section 4.3, along with technical approaches or solutions for mitigating unacceptable performance. In this section, the key technical issues which have been identified as unique to the proposed 35x35 MHz band plan are described, and approaches to resolve or mitigate these issues are presented. In some cases the resolution approach entails use of interim viable solutions with a transition plan, as applicable, to take advantage of expected technology advances for more optimal longer term solutions. Planning for reasonably expected technology advances allows a band plan to be selected that is more efficient and effective than one based on a shorter term viewpoint that is strictly limited to currently available technology. A forward-looking approach will provide the maximum value to end users and operators. The technical issues identified as posing a unique challenge for the 35x35 MHz plan include (1) antenna performance, which is closely related to user device size and cost; (2) RF duplex filter feasibility and performance; and (3) interference generated to other LTE bands operations due to transmitter harmonics. Each of these issues is addressed below : Antenna Design Requirements One of the key requirements for supporting a cost effective multi-band LTE handset design is minimizing the number of physical antennas to fit within the user device form factor-- ideally it is desirable to have a single antenna for bands below 1 GHz. This implies the use of antenna structures with a tuning approach to optimize performance for a specific band. FINAL 14 3/11/2013

127 Primary antenna design parameters include its length and the dimensions of the ground plane in the device. The primary (external) antenna performance characteristic is its RF radiation efficiency. The antenna length is dependent on the center frequency of operation and total range of operation. Antenna size is restricted to the space available in the user device. Internal antennas are a requirement for a competitive consumer handset (smartphone) form factor. Antenna efficiency in a specific band of operation is driven primarily by the size of the pass band and the center frequency. The pass band requirement arises since the antenna has to operate on both the uplink and the downlink. The ratio of the size of the pass band to the center frequency determines the 1 db efficiency parameter [QUAL1]. The Operating Spectrum and its relationship with the center frequency of operation also affect antenna efficiency. Antenna is optimized for minimal reflection coefficient and impedance matching at its input and output ports, and low cross coupling between the ports. Antenna RF efficiency, from the standpoint of maximizing received or transmit power, is typically focused on performance optimization for the uplink due to limited UE power, at the expense of downlink performance, where high LTE base station power is easier to provide and can compensate for user device inefficiencies. Figure 3 provides a conceptual (and highly simplified) reference for the discussion of the device antenna design parameters and performance discussion. Figure 3 Reference Diagram for Device Antenna In this diagram, the device size footprint (dashed box) is shown encompassing the antenna with length L and the associated ground plane which determine the RF performance, including frequency of operation and bandwidth. The three (potential) matching networks for three corresponding bands of operation are shown schematically, along with a potential per-band tuning approach for the antenna itself. Antenna Size for 35x35 MHz Band Plan Using Traditional Passive Antennas Traditional passive antenna implementations have used dipole and monopole implementations. Based on electromagnetic resonance phenomena, the half wavelength of the center frequency determines the required basic antenna length for traditional passive antennas. This corresponds to 228 mm length for the center frequency of 658 MHz associated with the 35x35 MHz band plan. Also the total range of bandwidth from 1 GHz to cover the 35x35 MHz band plan becomes 382 MHz FINAL 15 3/11/2013

128 which drives the incremental length increase and affects antenna efficiency. Table 1 provides the impact of the operation at 35x35 MHz band plan in reference to the 25x25 MHz band plan [ATT1]. 2 Table 1: Antenna Length Using the contributions of the center frequencies and the wider band width associated with the 35x35 MHz band plan, the estimated first order increase in the length of the antenna is 6.9% as compared to the one for 25x25 MHz operation. In case the additional length of 6.9% as compared to the 25x25 MHz cannot be accommodated and the shorter 225 mm antenna needs to be used, then the loss of receiver sensitivity is estimated to be db. Such a comparatively low level of detuning can be mitigated by correspondingly increasing the transmit power. Concerning transmit power, in normal operations, the User Equipment (UE) power is being adjusted dynamically. In particular, as the UE gets closer to the base station, the power is reduced. The UE s which generally transmit the maximum power of 23 dbm are near the cell edge. Hence it is possible to increase the transmit power for a significant number of the UE s. However, most practical systems also use smaller antennas as indicated in the footnote, and hence the issues associated with use of half wavelength antennas are diffused. Alternatives including active antennas, printed antennas etc. already exist in the marketplace to handle the requirements for the 35x35 MHz band plan very effectively without loss of performance. The size of such advanced antennas is generally half of that of the traditional passive antenna. Antenna Efficiency To-date, since the lowest frequency of LTE operation has been in the 700 MHz band, it becomes desirable to provide commonality for operation with a 700 MHz design. 700 MHz antenna design 2 In an actual smartphone design, the antenna length will be scaled (down) in order to meet the requirement for an internal antenna and the associated ground plane, and a tuning network may be introduced to optimize performance at the specific bands of operation. For example, a 5 inch long smartphone may only be able to accommodate an antenna of length 60 mm. The relative comparisons in the table would still be valid for the actual (scaled) antenna. FINAL 16 3/11/2013

129 provides a reference point for the performance of a 600 MHz antenna. The various tuning options and the applicability of antennas for 700 MHz and 600 MHz operations are summarized in Table 2: Table 2: Antenna Tuning Options As indicated in [QUAL1], a 700 MHz antenna with a center frequency of 710 MHz is tuned and optimized for 6%, 1 db efficiency 3. The antenna efficiency decreases as 1/f 3 where f is the center frequency. If the same physical antenna is used for the 35x35 MHz operation, the resulting detuning (less efficient operation) results in 1 db efficiency of 4.6% as compared to the 5.3% requirement for the 35x35 MHz band plan [QUAL1]. The estimated detuned impact of using a 700 MHz tuned antenna for the 35x35 MHz band plan is -0.60dB. Such detuning at 600 MHz can be overcome by using active components for tuning [MAKI]. Another approach is to retune the 700 MHz antenna to a 6.9%, 1 db efficiency for 700 MHz band operation so that the same tuned antenna can be used for the 35x35 MHz operation. This would be consistent with the desire to reuse the 700 MHz antenna for 600 MHz operation. A third option is to use an antenna with the increased length for 600 MHz operation and tune it to the 5.3% requirement for the 35x35 MHz band plan. A single antenna is acceptable for all frequency bands so long as all the antennas can be arranged within the same space of the physical limitation of the lowest resonant antenna [SONG]. The above analysis shows that using traditional dipole antennas (or scaled antennas) even if tuned to the 700 MHz band, when used for 600 MHz bands will have marginal external performance degradation. Concerning the feasibility of supporting the operating spectrum of 80 MHz, current designs can already support 65 MHz [VERI] and it is expected that by the end 2014 / early 2015 timeframe, antennas supporting operating spectrum of 80 MHz will be available based on the discussion in the next section. This is consistent with the requirement for the 35x35 MHz band plan. Technology Feasibility of New Antennas for 35x35 MHz Band Plan Many recent technology advances have made it possible to address the performance limitations and size issues associated with traditional passive antennas by utilizing microstrip and active antenna technologies [GUHA], [SHMB]. These have been designed with the objectives of covering a range of technologies which include LTE band plans along with the existing 2G systems, 3G systems, and Wi- 3 This implies that the filter roll-off of -1dB from the center maximum will correspond to a bandwidth of 42.6 MHz (6% of 710 MHz). FINAL 17 3/11/2013

130 Fi and Near Field Communication systems [MOBI]. In addition to broadband antennas covering the LTE bands, a front-end chip set to support the wide LTE spectrum range has also been announced recently [QUAL2]. Antennas for multi mode, multi frequency, and other non-cellular technologies are based on several advances which include isolated Mode Antenna Technology (imat) [SKYC], surface mounted band switching and active impedance matching [SHMB], printed loop antennas [WONG], ceramic substrates [MAKI], and helical antennas [EGOR], among others. These antennas are also being designed to support Carrier Aggregation (CA) and MIMO systems. The primary focus of the antennas designed to date has been to accommodate the recent 700 MHz band plans; however, the same techniques can be extended further to accommodate the proposed 600 MHz plan. For example, a surface mounted antenna covers the 690MHz to 2.7 GHz and is only 3 inches in size [MOBI]. Some of the recent implementations and their possible applicability to the 600 MHz band usage are described below. A tunable antenna [SKYC] is based on the isolated Mode Antenna Technology (imat) and uses less than half the volume of traditional antennas. This enables a single and compact antenna structure that can operate on up to 12 transmit and receive bands along with MIMO technology. Another effective approach uses band switching and active impedance matching. Such antennas have already been introduced in commercial devices [SHMB]. This approach supports 13 bands and enables a smaller, more resonant antenna, providing dynamic tuning across a wide frequency range. Band switching enables the antenna to be half the size of traditional antenna to fit in the thinnest smartphone devices. In addition, innovations leveraging active impedance matching are expected to continue to be developed, which will allow the antenna to be matched dynamically and also enable quick adjustment for re-matching the antenna across wide bandwidths. In order to reduce the size of the antenna assembly for small user devices, a compact integrated dual port antenna has been reported [RAO]. This merges two inverted F Shaped Antennas (PIFAs) into a single antenna structure and such an antenna has already been introduced in commercial products [RAO]. For LTE operations in the wide range of spectrum from 700 MHz to 2690 MHz, use of a small sized printed loop antenna with two strip monopoles for multi band operations has been proposed [WONG]. An enhancement to this approach using a folded loop with a capacitively coupled feed covers the spectrum below 1 GHz in a compact size of 60x10x6.5 mm [CHIU]. The primary issue in planar antennas as compared to dipoles is their interaction with the device ground plane. One standard mitigation approach is the physical separation of the planar antenna from the ground plane inside the user device. These techniques illustrate how wideband operation at 600 MHz can be supported effectively based on currently available, advanced commercial products within the smaller dimensions required for a smartphone. Supplementary Downlink and its Impact on Antenna It is further noted that Supplementary Downlink (SDL) spectrum is being actively considered for all band plans. The primary technique for using SDL is expected to be carrier aggregation. Independent of the initial size of the band plan, whether 25x25 MHz or 35x35 MHz, it becomes necessary to expand the antenna length to handle the expanded operating spectrum range which includes the additional SDL spectrum, or otherwise accommodate the performance degradation that accompanies the wider bandwidth of operation. In essence, very similar impact occurs on the FINAL 18 3/11/2013

131 antenna length and efficiency regardless of which core band plan (35x35 MHz or 25x25 MHz) is chosen. Since all band plans augmented with SDL use essentially similar overall operating spectrum range, they will all require use of the same antenna subject to the accommodation of the pass band differences. In case a range of 120 MHz total operating range is assigned including uplink, duplex gap, and the paired and unpaired downlinks, then regardless of the band plan chosen, the operating bandwidth becomes approximately 18% of the center frequency. For efficient operations, a common transmit antenna may be used for both the narrower 700 MHz and 600 MHz uplink operations. On the receive side, a separate tuned antenna for the 600 MHz operation may be used to accommodate increased bandwidth due to SDL and to improve downlink efficiency : Duplex Filter Design In a user device (handset or smartphone) the RF duplex filter is the component in the path between the transmitter or receiver amplifier and the antenna. (see Figure 4) Figure 4: Duplex Filter It prevents the relatively higher RF power produced by the device transmitter (TX) from entering the device receiver (RX) which operates at much lower receive power levels. The primary technologies currently used for cellular operations are the Surface Acoustic Wave (SAW) and the Film Bulk Acoustic Resonator (FBAR) technologies [ALU]. Similar to the Antenna tuning issue, filter implementations are driven by the ratio of the pass band to the center frequency of operations. The current state of the art allows the ratio to be in 4% range [QUAL1]. Note that the corresponding figure for Band 3 AWS is 4.2% [3GPP]. This implies the currently supported pass band to be 28 MHz for the 600 MHz operations and it is estimated that technologies in the end 2014/early 2015 timeframe will support a pass band of 30 MHz [ALU], [QUAL1]. In case filter design limitations will not support the proposed pass band of 35 MHz with 10 MHz duplex gap separation, a commonly used approach [ALU], [QUAL1] is to split the pass band into two regions and use two filters to support the segments of the overall 35 MHz. The segmentation should be done on a 5 MHz block boundary in the uplink and downlink. A representative segmentation is shown in Figure 5. FINAL 19 3/11/2013

132 Figure 5: Filter Segmentation Illustration Segmenting the pass band into two filter segments may also have some advantages in certain situations where there is interference, e.g., due to TV operations, in part of the total range. Use of two filters provides some protection from interference due to operations in the other segment, protection that a single filter may not be able to provide. A practical implementation of a segmented filter may require two filters covering 2/3 rd of the overall bandwidth with overlap [VERI]. Each of these filters may have 25 MHz bandwidth and the overlap between them provides good characteristics for channels of up to 20 MHz size (see Figure 6). Figure 6: Two 25 MHz Overlapping Filters The filters require associated switches, but the rest of the hardware including amplifiers and antenna can be common. New technologies like silicon on sapphire for switch paths and digital capacitors and tunable filter banks [MAKI] result in manageable additional cost and space impact to accommodate the two filters. The filters can be tuned and configured via software. A typical size of a single filter is 8 mm 3 [EPCO]. Having a second filter along with additional glue (interconnection) hardware may result in total size of 20 mm 3 which is relatively insignificant compared to the size of FINAL 20 3/11/2013

133 a typical user device. A compact dual filter packaging has already been shown to offer space and cost savings [MAKI]. As technology advances enable the availability of a single efficient filter in the 2015 timeframe, the two filters can be replaced by a single filter. Continued advances in both SAW and FBAR are being done especially in packaging [MAKI}. Adaptive matching using tunable capacitors will allow multiple bands to be handled with significant power savings [MAKI]]. Hence a very effective migration path exists from the two filter interim implementation to more optimal (smaller and lower cost) single filter implementation as technology evolves. The two filter interim solution has major advantages as compared to the two band approach considered in some of the proposals to address the issue of a wide pass band. A single band approach with two internal filters proposed here provides economy of scale, more efficient interoperability, common antenna design, less network management requirements, and a migration path which leverages technology advances. It may be worthwhile to mention here that a wider FDD based 45x45 MHz band plan formulated in 2010 has already been proposed for harmonized operations in the MHz spectrum for the Asia-Pacific region [APT]. Similar to the 35x35 MHz band plan under consideration here, it uses dual-duplexer filter arrangement to facilitate mobile user device implementation with the overlap to provide flexibility to administrations in their national spectrum planning. The proposed duplex gap is also 10 MHz with the key parameters as 6.0% (pass band / center frequency) and 13.2 % (operating spectrum / center frequency). These are more aggressive than the corresponding ones for the 35x35 MHz band, i.e., 5.3% and 12% respectively. Consistent with the discussion in Section on the implications of using Supplementary Downlink with any band plan, the addition of SDL will require antennas and filters to be extended to support operating spectrum frequency of approx 120 MHz down from 698 MHz. Hence the antenna and filter advances in areas of active tuning, surface mounted and PCB implementations etc. will clearly facilitate the resolution of the antenna and filter issues relating to the 35x35 MHz band plan : Harmonic Interference Background All RF transmitter amplifiers used in wireless devices (smartphones) generate undesired signals due to non-ideal (non-linear) transmitter operation in the uplink. It has been shown that harmonics generated from base station emissions are manageable [ALU]. Sound design will always minimize the unintended emissions from user devices that are generated along with the desired transmit signal. The undesired signals are created at harmonics (integer multiples) of the desired signal frequency [GHAR]. The undesired harmonic signals are created at power levels that are significantly lower than the power level of the desired signal, and decrease rapidly as the harmonic (multiple of the desired frequency) increases. These depend strongly on the non-linearities of the amplifiers and other elements. Typical values for the relative harmonic power levels due to such non-linearities are plotted in Figure 7. FINAL 21 3/11/2013

134 Figure 7: Illustrative Power Levels of Transmitter Harmonics With a typical maximum value of 23.5 dbm for the desired transmitted signal for an LTE user device, the power levels of the second and third harmonics (frequency multiples) are approximately at -48 dbm and -70 dbm respectively, at the device output. The 4 th and 5 th harmonics are in the -80 dbm range, and subsequent harmonics are in the -85dBm to -100 dbm levels. For the 35x35 MHz band under consideration, the affected spectrum pertaining to the various harmonic frequencies is indicated in Table 3. Table 3: Harmonics for the MHz Uplink Transmission Harmonic Frequency range (MHz) Harmonic Frequency range (MHz) 1 st th nd th rd th th th th Harmonic interference can occur when the undesired harmonic output of user device at the higher frequency leaks over to the same user device, or another device in close proximity, which is receiving at the higher frequency simultaneously on the downlink. In addition to degrading the performance (throughput) of the second device, this interference may also impair the advantages of carrier aggregation. FINAL 22 3/11/2013

135 Analysis of Harmonic Interference The second harmonic of the 35x35 MHz band plan uplink occurs in a spectrum region which does not have active operations, and hence its impact need not be considered. The third harmonic created by the 2 MHz frequency range MHz in the uplink lies in a 1 MHz overlap region common to both the PCS Band 2 and Band 25 ( MHz). The third harmonic may also affect an additional 5 MHz in the PCS Band 25 ( MHz) as shown in Figure 8. Figure 8: Harmonic Interference into the PCS Band It is estimated that at its maximum power output of 200 mw (23 dbm), a 600 MHz device (uplink) transmitter in the user device will inject -26 dbm in a PCS receiver in a nearby device [ALU]. Assuming some level of reasonable mitigation and based on the results from the corresponding scenario between 700 MHz operation and its third harmonic interference into the AWS band [ATT1], this is estimated to create an, approximately 7 db reduction in the PCS antenna or correspondingly, in receiver sensitivity. Typically, in communication systems, a 3dB signal loss may correspond to loss of 1 bit/symbol. For LTE using a symbol size of 6 bits for 64 QAM operation, this maps into throughput loss of 40% over the frequency spectrum affected. This is a worst case condition, as the typical output power of a user device may be much lower than 200 mw. It is also noted that power in the MHz segment causing the interference is shaped (reduced) according to the standard 5 MHz LTE block which has 4.5 MHz operating band with 0.25 MHz as internal guard bands on both sides [3GPP]. Consequently, the interference caused in the PCS band is further reduced (see Figure 9). FINAL 23 3/11/2013

136 Figure 9: Third Harmonic Interference Due to LTE Block Characteristics For the PCS band 25, this implies a capacity loss of 3.5% and correspondingly 0.42% for the PCS Band 2. These are relatively low levels of loss, again calculated at worst case power output levels and can be mitigated as indicated below Additionally, this type of interference will occur only when devices which are transmitting in the 600 MHz band, are also simultaneously receiving in the PCS band, or in close proximity to a device receiving in the PCS band. It is also important to observe that the adverse effect of capacity loss for users can be curtailed by ensuring that dropped calls are minimized, as is currently done in cellular systems. Preference is given to existing sessions instead of allowing new sessions to come in. This is part of considering existing calls to be at a higher priority than new calls as part of the Key Performance Indicator (KPI) management. Hence generally, unless extensive additional interference occurs as a mobile user moves, reduced capacity will not manifest itself as dropped calls. The relatively low levels of capacity loss estimated at worst case conditions due to harmonic interference can be further reduced or eliminated with sound engineering. Options for hardware solution include making the amplifier more linear, creating better isolation between the 600 MHz transmitter and the PCS receiver components, using different linear polarization orientations for the two antennas, or adding a harmonic filter to suppress the 3 rd harmonic. A Harmonic rejection patch antenna has also been proposed for suppressing 2 nd and 3 rd harmonics very effectively [BINM]. A nominal isolation level of 7 db based on use of the harmonic filter and the harmonic rejection antenna can negate harmonic interference, Patch antennas based on printed circuit technology are very compact with low cost and high reliability. Examples of use of multi pole elliptical filters to suppress higher levels of harmonics have also been reported [AGIL]. These solutions are expected to have manageable impact on space and power requirements in the user device. Additional mitigation could involve coordination or exchange of spectrum between operators with AWS and PCS holdings [NSN]. FINAL 24 3/11/2013

137 Typical criteria for severity of harmonics have been proposed in [QUAL1]. It is indicated that a harmonic level below -80 dbm may be considered to be of low severity. As seen from Figure 7, the levels for 4 th and higher levels are generally in the -80 dbm or below. Hence their impact can be neglected (see also [ALU]. In particular, the 4 th harmonic of the 35x35 MHz uplink band plan may potentially interfere with the Educational Broadband Service/Broadband Radio Service (EBS/BRS) band, but its effect is reduced an additional 10 db as compared to the 3 rd harmonic, resulting in a signal level below the typical sensitivity of -100 dbm [ALU], [3GPP]. Therefore, this is not expected to compromise international harmonization where the BRS band is being deployed. If necessary, mitigation approaches similar to the ones mentioned above can be applied. It may be noted that it is not sufficient that the appropriate harmonic of an operating uplink spectrum fall in the downlink range of a potential victim to cause interference but the interference level should be sufficient to cause measureable adverse impact on the potential victim. 4.3: Issues Common to All Major Proposed Band Plans The key technical issues which have been determined to be in common for all the major band plan proposals for the 600 MHz spectrum, including the 35x35 MHz band plan, are described in this section, along with technical approaches or solutions for mitigating adverse performance effects. The key issues include the following: The size of the Duplex Gap (DG) and the types of applications allowed in the that gap Interference due to TV operations in the uplink domain or adjacent to downlink lower edge in a geographically overlapped region. It may be noted that one of the fundamental principles is that TV operation is not allowed in the downlink region itself. The Guard Band (GB) to protect the cellular operation from the TV operation and vice versa, for operations in adjacent spectrum in an geographically overlapped region Interference due to TV operation in the same channel in geographically adjoining and national boundary regions Interference due to TV operation in an adjacent channel in geographically adjoining and national boundary regions The following sub-sections provide the framework, the impact, and the mitigation approaches for these issues applicable to all of the band plans : Duplex Gap For Frequency Division Duplex (FDD) operation, a Duplex Gap is required to isolate the uplink (UL) spectrum from the downlink (DL) spectrum. For practical systems, there needs to be a gap between the uplink and the downlink band for a filter to be able to provide isolation between the device (handset or smartphone) transmitter and receiver. Experience has shown that a duplex gap of approximately 1.5% of the center frequency [ALU] allows for a practical duplex filter design. In the case of the 35x35 MHz band plan with a center frequency of 658 MHz, this experience implies a FINAL 25 3/11/2013

138 minimum duplex gap of 9.9 MHz. Hence for the various band plans, the range of MHz is recommended [QUAL1]. It may noted that there is tradeoff between increasing the size of the duplex gap for ease of filter design and decreasing it in order to improve the spectrum usage efficiency, since the duplex gap represents spectrum unusable to the wireless system. The size of duplex gap in current 3GPP band plans varies from 13 MHz (1.8% of center frequency) to 355 MHz (18.4% of center frequency) [3GPP]. It may be noted that in many LTE band plans, the duplex gap may be created by locating other LTE band plans in the intervening space or using the duplex gap for Supplementary Downlink (SDL) and other non-interfering applications. For the situations like the 600 MHz spectrum with uplink and downlink pass bands next to each other, it is appropriate to use the reasonable duplex gap size of 10 MHz. This is also consistent with international harmonization efforts. If the Duplex gap size is of relatively narrow bandwidth, the impact of the duplex gap size on the antenna is minimal if the gap is contiguous with the downlink and uplink, and it is also roughly independent of the width of the uplink or downlink pass band. The increase in the overall bandwidth resulting from a much wider duplex gap has a corresponding impact on Antenna size, as it increases the required operating range of the antenna. It is also important that the duplex gap be aligned for all operators for nationwide deployment, enabling use of a single device filter and low cost user device design. In order to recover the spectrum usage efficiency in the presence of the duplex gap, non-interfering networks may operate within it. It is critical that the application in the duplex gap not interfere with the uplink and downlink operations. The duplex gap may be used for applications like (low power) wireless microphones [QUAL1] or for LTE Supplementary Downlink (SDL). The SDL would be used as a supplemental downlink frequency for uplink bands other than the upper 600 MHz uplink under consideration here. Use of Guard Band for Broadcast TV Operation In case the use of the duplex gap is being considered for TV operation, then corresponding guard bands needs to be placed on both sides of the 6 MHz TV band after appropriate alignment with the uplink and downlink boundaries to minimize interference. This implies total of 26 MHz guard band between the uplink and downlink with 20 MHz of the spectrum being lost. Hence insertion of TV operations in the duplex gap is not recommended. Reducing the size of the guard band would cause a TV-to-broadband wireless interference potential : Interference due to Broadcast TV Operations The interference between the TV transmission and the proposed cellular operations in the 600 MHz bands can manifest itself in the following three major ways: 1. Inter-Modulation Interference (IMI) 2. Co-Channel Interference (CCI) 3. Adjacent Channel Interference (ACI) Interference depends upon several factors including the TV and cellular base station power, distance and distribution of the LTE user devices and the base stations, the TV transmitter and receiver distributions, signal sensitivities of the user device and the TV receiver, geographical FINAL 26 3/11/2013

139 overlaps and / or adjoining placements, propagation path characteristics, and the TV and LTE antenna and filter characteristics etc. Hence interference and its impact is complex to analyze and model and the results can vary significantly depending upon assumptions and variability in various parameters mentioned above. In this study, rough estimates of the impact of interference are derived based on measurement reports, high level modeling, and use of extrapolations and interpolations. It may also be worthwhile to mention that whereas an attempt is made in this study to segment the various interference scenarios into distinct combinations of spectrum adjacency and geographical co-location and separation, these are not necessarily mutually exclusive. In practical systems, the real interference is likely to be weighted combinations of these. The primary intent is to capture the expected impact on performance and identify possible mitigations. In actual implementations, the impact and the mitigations may need to be re-evaluated. Typically, the moderate power TV broadcasts can be in the 50 KW range and the high powered ones in the 1 MW range. TV transmitter coverage may extend to large metropolitan areas. Base stations transmitted power may range from 50 W to 1 KW and the base stations may typically be located ¼ to 4 miles apart. The 3GPP standards specify the cell phone transmit power to be between -40 dbm and 23 dbm (200 mw) [3GPP]. Based on the distances between base stations, the received signal strength at a user device may be in the range of -30 to -80 dbm [QUAL2]. Inter-Modulation Interference Inter-Modulation Interference (IMI) arises when the TV signal interacts with the uplink frequency signal from a user device and creates inter-modulation products. The interference occurs at sums and differences and multiples of those sums and differences of the two interacting frequencies. These inter-modulation products create possible interference in the receiving downlink segment of a cellular device. The initially proposed FCC band plan in the NPRM document [FCC] indicated the possibility of having TV stations in the duplex gap. Several submissions to FCC in response to the FCC NPRM document have pointed out the adverse effect of IMI in the downlink of the band plan. The currently proposed band plans, including the 35x35 MHz being considered here recommend not using TV transmission in the duplex area which could create such IMI. Hence IMI is not considered further in this study. Co-Channel and Adjacent Channel Interference Co-Channel Interference (CCI) is the result of interactions between the TV signal and the cellular signal (uplink or downlink) in the same channel. This generally creates reduced Signal to Noise Ratio (SNR) to the base station, the cell phone, or the TV receiver with resulting impact on throughput and capacity of the cellular and TV networks. Adjacent Channel Interference (ACI) is similar to co-channel interference except for the operating frequencies of the TV operation and the cellular operation being in adjacent channels instead of being in the same channel. The interference issues being considered here are very similar to the ones in the operation of the 700 MHz band. [SUPT]. Significant progress achieved in mitigating the interference in the 700 MHz band region is directly applicable to the 600 MHz operation. LTE has mechanisms to handle interference and reduced signal to noise ratio by employing lower modulation levels which reduce the data rate and throughput, but maintain the desired Bit Error Rate (BER), a required part of meeting the Quality of Service (QOS) requirements in an IP system. LTE supports three levels of modulation: 64 Quadrature Amplitude Modulation (QAM), 16 QAM, and Quadrature Phase Shift Keying (QPSK). Taking the 64 QAM throughput as a reference (BW ref), FINAL 27 3/11/2013

140 the corresponding throughput is typically BW ref / 2 and BW ref / 4 for the 16 QAM and QPSK respectively (see e.g., [GUID]). The decreased SNR may also entail packet error management mechanisms which may result in re-transmission of packets between the user device and the base station. There is hence a corresponding reduction in throughput as SNR is reduced due to interference. If SNR goes below a pre-defined threshold which does not support QPSK, then the user may need to be dropped. Such a situation results in reduced capacity. For a typical distribution of users across the whole cell coverage range, the capacity decrease may be estimated to be on the order of 1/3 of the maximum throughput. It is expected that TV operations will be vacated from the complete downlink region above channel 37 for nationwide consistency and commonality of user devices. However, depending upon the outcome of the reverse auction, TV operations may still exist in the uplink domain in some selected regions of the country. Also, TV operations may continue in Canada and Mexico creating interference near the international boundaries. This entails the possibility of various types of interference scenarios with the cellular operations. The relationship between the type of interference and the geographic adjacency including border areas is summarized in Figure 10. Figure 10: TV Co-Channel and Adjacent Channel Interference FINAL 28 3/11/2013

141 The Interference scenarios for Uplink and Downlink are summarized in Table 4. Table 4: TV Interference Scenarios* Note that overall interference may be combination of overlapped and adjoining interferences depending upon the actual geographic relationship between the TV and cellular operations. Similarly, overall interference may also need to take into account the possible combination of the channel overlap and the adjacency while calculating interference. Geographically Overlapped Region First consider the case of the geographically overlapped region. Clearly due to interference considerations, cellular and TV operations must not coexist in the same channels. Adjacent channel operations can, however, occur (Case A in Figure 10). The spectrum location of the TV implies two cases. The first case is the operation of the TV station operating in the uplink domain of an upper 600 MHz band plan ( MHz). In this case, the cellular uplink operation is vacated from the channel(s) that the TV station is using and moved to an adjacent channel. The other case corresponds to the operation of a TV station below the lowest edge of the downlink (below 608 MHz). Both situations correspond to operation of cellular system in an adjacent channel to the TV operation. Adjoining and Border Areas In adjoining and border areas, TV transmission may be in the same channel as the cellular operation or in an adjacent channel. Both co-channel and adjacent channel interference are applicable. The one exception is that TV interference in the downlink channels will not occur inside the US boundary since TV operations in downlink spectrum are not planned to exist in US. However, such interference can still arise on the national boundaries with Canada and Mexico. FINAL 29 3/11/2013

142 The interference victims for uplink and downlink signals are shown in Figure 11. Figure 11: Targets of Potential Interference The victim entities impacted by such interference are summarized in Table 5. Table 5: Interference Victims In the case of the TV operation being in the cellular uplink region operating in a channel adjacent to a cellular uplink block, the cellular base station and the TV receiver are the potential interference victims. The cellular base station may lose capacity. The TV receiver in the vicinity of a cellular user device may have degraded operation. This may result in coverage loss for the TV broadcast system. FINAL 30 3/11/2013

143 It is acknowledged that it is quite difficult to predict the impact on a TV receiver due to a mobile user device [SAML]. In the case of the TV operation being in a channel adjacent to the downlink band or in the downlink spectrum, the cellular device and the TV receiver are the potential interference victims. The cellular user device may experience loss of data and dropped sessions and overall, the capacity of the cellular system may suffer. The TV receiver situation is similar to the previous case of uplink : Geographically Overlapped Region Adjacent Channel Interference This situation corresponds to the Scenario A in Figure 10 where the TV and LTE networks are expected to operate in adjacent bands in the same geographically overlapped area. This necessitates the insertion of a guard band between the TV channel and the cellular band. The tradeoff for the decision on the size of the guard band is between capacity loss due to interference which decreases with increasing guard band size and overall spectrum availability which increases with decreasing guard band size. TV Operation in the Uplink Range Figure 12 shows the spectrum overlapping region between the roll-off segments of the spectrum characteristics of the TV band and the adjacent LTE band. Figure 12: Guard Band for Minimizing TV Interference For the case of the TV operation in the uplink domain and the cellular system uplink being in an adjacent block, the associated capacity loss due to interference with the base station can be mitigated using guard bands. The level of interference is clearly dependent on the size of the guard band. Several measurement studies have been carried out to capture the various interference FINAL 31 3/11/2013

144 scenarios. One prominent study [CEPT] identifies the impact on the cellular base station due to the TV broadcast in an adjacent channel. In case of no guard band, the impact can be as high as db whereas use of guard band of MHz results in minimal out of band blocking level. The interference on a typical TV receiver due to a nearby cellular uplink device in an adjacent channel without a guard band would have a TV coverage loss of 5-10% in case the cell phone and the TV receiver are in immediate vicinity [RNDH]. This can be mitigated by increasing the TV power by 0.5 db. It may be noted that since an LTE user device is comparatively low transmit power device, the interference into a TV receiver is minimized for most practical situations when the distance is large or there are signal blocking entities, e.g., indoor / outdoor operations. In addition, significant numbers of indoor TV receivers receive their signals through cable/fiber or through satellite signals. LTE user device uplink signals will not interfere in such situations especially if the cables and other sensitive hardware are shielded. Another study estimates that there is minimal interference to the TV receiver from a cellular user device in case of an 8 MHz guard band [CHO]. Hence in case of TV operation in the uplink range, a guard band of 10 MHz is recommended with respect to the adjacent cellular band due to the impact on cellular system [ALU]. This is particularly important so as to decrease the impact due to TV operations on any of the uplink blocks. This would imply that all the blocks will essentially be technically equivalent. For moderate power (50 KW) TV stations, a guard band of 6 MHz is considered acceptable [ATT1] The 10 MHz guard band in between the LTE operations and a high power TV station has also been the route the 700 MHz community has been taking as illustrated by the design of the commercial guard band filter used at the low end of the spectrum [ALU] [EPCO]. The frequency response of the commercial EPCOS filter for the 3GPP band 12 in the lower 700 MHz region to reduce interference from the TV channel 49 is reproduced in Figure 13. Figure 13: 10 MHz Guard Band between Band 12 and TV Channel 49 [ALU] FINAL 32 3/11/2013

145 TV Operation near the Downlink Cell Edge The situation of the TV operation next to the downlink of a UMTS system has also been studied previously [CEPT]. This analysis is expected to apply to LTE operations. The loss of capacity is characterized in two environments at the annular area at the cell edge (see Figure 14), and as an average across the whole cell area. Figure 14: Cell Edge Area The results for the cell edge capacity loss are summarized in Table 6. Table 6: Cellular Capacity Loss near Cell Edge Guard Band Cellular Capacity Loss (%) No guard band 24% (extrapolated) 3 MHz 20% 8 MHz 14% It may be noted that when the impact of interference is averaged over the whole cell, then the capacity loss is 5% without a guard band. The corresponding impact on the TV coverage even with a 1 MHz guard band is estimated to be 3% [CLEA]. Similar to the case of the TV operation in the uplink regions, in the case of TV operation below the Downlink edge, a guard band is expected to address the interference issue. For the proposed 35x35 MHz band plan, a 10 MHz guard band is created by inserting a 4 MHz Guard band which augments existing isolation provided by the 6 MHz channel 37. Channel 37 applications operate in a noninterfering mode with the LTE bands. Thus the 35x35 MHz band plan provides a very efficient usage of the spectrum above channel 37. FINAL 33 3/11/2013

146 In case there is a need to further reduce interference after the introduction of a guard band, or if adequate interference reduction has not occurred in specific cases, then several additional measures have been recommended [CLEA] as summarized in Table 7. Table 7: Additional Mitigation Approaches Unit Mitigation Applicability TV near Cellular UL TV near Cellular DL TV Receiver In Line Filter X X Signal Selectivity X X TV Transmitter On-Channel Repeaters X Antenna Height Antenna Pattern and Beam Forming Co-siting (with cellular base station) X x x Cellular Base Station Transmitter Filtering X Orthogonal Polarization Power Reduction Antenna Height Antenna Pattern and Beam Forming X X X x The In-line filter option for TV receivers is considered impractical because of the necessity of upgrading all affected customer TV receivers. The on-channel Repeater approach basically involves increasing the transmitted power of a TV station. This is most effective of the various solutions but also is most expensive and may also add interference to the cellular user device. Adjustments of antenna characteristics (height, pattern, tilt, and direction) [SAML] are comparatively inexpensive and are part of finer optimization. This is true for the cellular base station as well. Co-siting and directing the energy for the TV station and the cellular base station is quite effective in situations where the respective coverage areas are comparatively disjoint. Base station transmitter filtering is also very effective and may cost approximately $600 per base station [CLEA]. Orthogonal polarization may give up to 16 db additional protection [SAML] and does not entail significant cost since it is a matter of orientation of the transmitted waves. The power reduction in the base station FINAL 34 3/11/2013

147 may be the least desirable option to reduce interference to the TV receiver since it reduces the cellular network coverage as well. Spectrum below Channel 37 It is also envisaged that the channels directly adjacent to and downward from channel 37 may not be used for TV transmission. Major alternatives exist for useful application of that range of spectrum, e.g., use as Supplementary Downlink (SDL) for the upper 600 MHz band plan or in support of the band plans in the 700 MHz, PCS, or other LTE bands. This will facilitate the support for many multimedia applications wherein the downlink requires substantially more bandwidth than the uplink. In fact, most technologies use the strategy of expanding the downlink with additional bandwidth. Similar strategy of enhancing the downlink performance applies in the access techniques used in downlink and uplink in LTE. LTE uses OFDMA for the downlink with its enhanced performance in contrast to SC-FDMA for the uplink. Hence in both the cases of geographically overlapped operations, whether the TV operation is in the uplink region or the downlink adjacent region, a guard band facilitates the mitigation of interference. This implies that the impact on the lowest downlink block is minimized and this makes the spectrum value of the lowest block similar to other blocks further away. Hence each block has similar propagation characteristics, similar interference features (including issues associated with interference from cross-border licensees in Canada and Mexico), and similar capabilities of communicating devices. This results in comparatively uniform value for the blocks in the entire spectrum range : Geographically Overlapped Region Guard Band The suggested guard band values are 6 MHz and 10 MHz for the moderate power 50 KW and the high power 1 MW TV transmitters respectively [RNDH]. It may be worthwhile to note that there is no specific threshold value for the size of the guard band. The primary trade-off is between decreased interference due to increase in the size of the guard band and the corresponding reduction in throughput and capacity. A typical 6 MHz TV station in the cellular uplink region may require guard bands on both sides unless it is at the low end close to the Duplex Gap. The size of the guard band also may need to be adjusted to align with the standard 5 MHz block boundaries for the LTE operation. A representative guard band scenario for a moderate power TV station operating in the uplink region is shown in Figure 15. FINAL 35 3/11/2013

148 Figure 15: Moderate Power TV Operation with Guard Bands A single moderate power TV station will result in a loss of 20 MHz of spectrum from the uplink region. As the number of TV stations increases, the corresponding adjusted guard bands may be needed between the end TV stations and the cellular operation. No guard bands are required between adjacent TV channels. The corresponding situation for a high power 1 MW TV station is depicted in Figure 16. Figure 16: High Power TV Operation with Guard Bands The reduction in the uplink region in the case of a high power 1 MW TV station is 25 MHz. To the first order, the primary impact of a TV station is to reduce the availability of cellular uplink blocks. It may be noted that the corresponding paired downlink block also becomes non - operational resulting in a loss of 5 MHz of bandwidth for each block. It is proposed that 3GPP consider as a future work item the feasibility of using a previously paired downlink block as a Supplementary Downlink in case the corresponding uplink is not available. It may be noted that TV operations cannot be inserted at the high end ( MHz) of the uplink spectrum because of possible interference with the low 700 MHz band. However, if the TV operations in uplink start at the low end (663 MHz of the 35x35 MHz band plan), then the duplex gap effectively acts as a guard band, and only one guard band is needed at the high end of the TV operation. Such typical situations are depicted in Figures 17 and 18 for the 25x25 MHz and the 35x35 MHz band plan respectively. FINAL 36 3/11/2013

149 Figure 17: Illustration for TV Operations in 25x25 MHz Uplink Figure 18: Illustration for TV Operations in 35x35 MHz Uplink It may be noted that in Figures 17 and 18, nominal guard bands of 8 MHz or higher sizes are being used so that the cellular blocks can be aligned with Multiples of 5 MHz boundaries. Table 8 summarizes the loss of throughput and hence the corresponding capacity level, due to elimination of the range of blocks due to the presence of TV stations from Figure 17 and 18. Table 8: Capacity Reduction for TV Stations in Uplink Region # TV Stations 25x25 MHz Band Plan 35x35 MHz Band Plan Available BW % of Maximum Available BW % of Maximum FINAL 37 3/11/2013

150 (MHz) Capacity (MHz) Capacity As indicated in Table 7, there is substantial loss of throughput and capacity in case of the operation of TV station(s) in the cellular uplink region. Because of the significant nationwide value for the cellular spectrum and the focus on providing the maximum spectrum to the customers, it is recommended that TV stations be relocated from the entire uplink spectrum nationwide for any of the band plans : Geographically Adjoining and Border Regions Co-Channel Interference This corresponds to the Region B in the Figure 10. The TV station, the cellular base station, and the user device may be located at arbitrary locations as shown in Figure 19. Figure 19: Geographically Adjoining Areas As stated earlier, the identification of interference and its impact varies according to the assumptions made for the transmit power, receiver sensitivity, differing coverage areas for the TV and the cellular operations, and geographic distributions. A geographically adjoining area implies the existence of TV station in one city or Economic Area (EA), with the cellular operations in a neighboring city or an EA. The border situation with Mexico and Canada is very similar to the one for adjoining area. A factor h may be used to indicate the fraction of the complete periphery relating to the adjoining interfering areas. As an initial assumption, the factor may be neglected and considered as 1. Co-channel Interference (CCI) will clearly be applicable in geographically adjoining and border areas. CCI arises from television stations operating in neighboring cities and towns in the same FINAL 38 3/11/2013

151 spectrum range that a cellular base station is using. The TV signal interference with the uplink signals from the mobile devices will impair the base station s ability to receive the mobile signals. This also affects the performance of the TV receiver. Even if all the TV stations in the US are removed from the cellular uplink spectrum and moved to other frequency regions, the scenario can still exist in the border region. Similarly, the interference between the TV transmission and downlink signal affects the performance of both the cellular phone and the TV receiver. This scenario is not applicable inside the US since all the TV stations are expected to vacate the downlink spectrum. The co-channel interference depends upon the overlap between the 6 MHz band of operation for the TV station and the 5 MHz block for the cellular system as shown in Figure MHz TV Spectrum 5 MHz LTE Spectrum Center Frequency Offset Figure 20: Co-Channel Spectrum Overlap It is clear from Figure 20 that such interference will occur in the spectrum overlapping region and will decrease as the offset between the center frequencies of the TV station and the cellular system increases. Extrapolating the measurements [ALU], [NSN], [CEPT] from the adjacent channel to a co-channel situation, the corresponding loss for a typical cellular system can be estimated. For typical TV broadcast, only the side lobes interfere with cellular systems in close proximity (< 1 Km). The placement of the TV antenna at large height and pointing of the main lobe at distances of Km results in interference which varies depending upon the distance between the TV station and the victim system. A typical value for such interfering signal may be -30 dbm [NSN2]. The measured signal strength due to a moderate power TV station at distances of km indicated -23 dbm [ALU]. [NSN2]. The corresponding signal is -10 dbm for a high power TV transmitter using a 13 db difference between the two. The signal is reduced further based on filter (-10 db) and antenna (-5 db) corrections at the receiving base station. This signal interferes with the signal received from a cellular user device. Extensive Monte-Carlo simulation results on the mutual co-channel interference have been published [GUID] with particular emphasis on interference by the TV station on the LTE base station in the uplink in the 800 MHz band. As a first FINAL 39 3/11/2013

152 order of approximation, the results are expected to be roughly applicable to the 600 MHz band region as well. Variations in several parameters have been studied including TV power and antenna patterns, TV duty cycle, base station distributions, It may be noted that the performance degradation also depends on the modulation scheme. Using typical values from the results in [GUID], high power TV operation may result in estimated 33% capacity loss. The variation of the capacity loss as a function of the offset between the center frequencies of the TV station and the cellular system using linear interpolation is depicted in Figure % % Capacity Loss (1 MW Transmitter) (Negligible Loss due to 50 KW Transmitter) Center Frequency Offset (MHz) Figure 21: Capacity Loss due to Co-Channel Interference The base station may be situated at different distances from the TV station. A first order approach to calculate the effect as a function of distance is to estimate the reduction in the interference signal strength and the corresponding reduction in the loss of capacity by noting that for free space propagation, the signal power decreases as power of two of the distance for omni-directional transmit antennas. For every doubling of distance, the power decreases by 6 db. Hence the loss of capacity also decreases correspondingly. This is indicated in Figure 22. It may be noted that this decrease in capacity loss may need to be modulated with the actual signal strength depending upon the TV antenna pattern as indicated above. FINAL 40 3/11/2013

153 Figure 22: Capacity Loss as a Function of Distance between TV Station and Base Station In this first order model, the ground path loss effect which depends on urban, semi-urban, or rural environments is not accounted for. If necessary, this can be done by including the corresponding path loss factor (see e.g., [HUFF]) which will reduce the capacity loss in this case correspondingly and hence interference as well. The primary approach for reducing and eliminating Co-Channel interference is to ensure that enough incentives exist for the TV station operator to vacate the whole uplink region for the 600 MHz spectrum. In case TV stations still exist, e.g., in the border areas, various approaches for interference mitigation or exclusion are feasible, and have been applied to 700 MHz operations [ALU]. These approaches include geographic isolation, co-siting and antenna orientation coordination, placing coordinated interferers on the same tower back to back, and use of ground level limitations of power density : Geographically Adjoining and Boundary Regions Adjacent Channel Interference The remaining C grids in Figure 10 also correspond to the cellular operation in a band in a region which is adjoining another region with TV operations. The TV channel does overlap the cellular system as in the case of CCI but it is adjacent to the one being used by the cellular system. Such interference arises due to the TV station transmitted power falling within the cellular spectrum due to the transmit filter roll-off, and can be considered to be equivalent to a typical adjacent band scenario considered in scenario A, but with the guard band being zero. The interference in this case is expected to be less as compared to the geographically overlapping scenario because of separation distance between the TV and the cellular coverage. As described previously, field measurements [CEPT] indicate that there is a 5% capacity loss across the cellular network due to the interference by the TV station. The further reduction in the capacity loss for adjoining and FINAL 41 3/11/2013

154 border regions will depend on the distance between the TV coverage and the cell coverage areas. Using the approach indicated in Figure 22, one can estimate that for a distance of 10 km, the loss may be as low as 1.2%. For both the co-channel interference and the adjacent channel interface cases, there are several mitigation techniques available [CEPT] (Also see Table 6). These including co-siting the respective antennas and pointing them in the appropriate areas, use of cross-polarization between the cellular system and the TV broadcast, using selective spectrum mask, adjusting power of the base station or the TV station, adjusting the antenna heights, beam forming the antenna pattern, installing rejection filters, and more stringent out of band emission control. Guard Bands have also been suggested to further reduce interference, [CEPT] but this results in a corresponding loss of throughput and capacity. In addition unlike a mobile user device, since the cellular base station is a fixed entity, network planning is also a tool for managing interference due to base stations. In all cases, it is undesirable to have separate market-by-market clearing targets for TV operation, with corresponding nationwide variations in the allowable spectrum for cellular operations : Interference Impact Summary As described in Sections to , various combinations of geographic separation, channel separation, and link direction result in different impact in the cellular and the TV networks. A summary of these is provided in Tables 9 and 10 for geographically overlapping regions and Adjoining / Border regions. Table 9: TV Operations in Geographically Overlapped Regions FINAL 42 3/11/2013

155 Table 10: TV Operations in Geographically Adjoining and Border Area Regions As Table 9 indicates, a major area of concern from a throughput and capacity viewpoint in geographically overlapped areas is allowing TV operations to exist in the uplink spectrum region ( MHz). The cellular operations in the channels being used by the TV clearly cannot be used by cellular systems in those markets. Cellular operations then need to be assigned adjacent channels. This requires at least a 10 MHz guard band in the useful uplink spectrum area and reduces correspondingly the blocks available for cellular systems. However, other situations, including impact on TV receivers, are minimal and manageable. One set of measurements has shown that the interference due to cellular uplink on a TV receiver is minimal if the LTE user device is more than 2.5 m. away from the TV receiver [RNDH]. In the case of TV operations in one area, cellular operations may be allowed in the geographically adjoining area in the same channel as TV or in an adjacent channel (see Table 10). The co-channel scenario is not applicable for the downlink within the US since all TV operations are removed from the cellular downlink spectrum channels. However, in case of border areas, TV operations may continue in the cellular downlink spectrum area and this will have major impact on cellular operations in US border areas. The adjacent channel operations are manageable with mitigation. It may also be noted that with the use of cable/fiber and satellite channels by TV receivers, the impact on TV reception due to interference from cellular operations is further minimized. However, TV broadcast operations in both uplink and downlink have significant impact on cellular operations. The risk of interference cannot be zero and judicious selection of mitigation techniques during implementation can result in acceptable performance for both the cellular and the TV operations [SAML]. These conclusions are applicable essentially to all the major band plans starting downward from 698 MHz. In all these cases, the higher frequency pass band is used for the uplink, and the lower for FINAL 43 3/11/2013

156 downlink. These are separated by a technically feasible and minimum sized Duplex Gap as discussed in Section Conclusions Using the core evaluation criteria, the analysis has identified the advantages as well as the technical challenges unique to the 35x35 MHz plan, compared to other plans. The primary advantage of this plan is the ability to provide additional 40% traffic capacity for users and operators as compared to a 25x25 MHz band plan. Unique challenges include handset antenna performance, duplex filter feasibility, and the potential for harmonic interference with a portion of the existing PCS band. The analysis further shows that current technology or anticipated technology improvements can overcome these challenges, and that even when these issues are taken into account, the 35x35 MHz plan provides the greatest amount of broadband wireless capacity to users and operators. In essence, instead of being limited by current technologies, the proposal exploits technological advances consistent with the 2014/2015 timeframe when the systems in the 600 MHz bands are likely to be introduced in the marketplace. These will result in more compact user devices and higher performance. This is considered as a more desirable and effective solution from a longer range viewpoint. Resolution or mitigation of potential technical issues unique to the 35x35 MHz band plan includes: Device Antenna: Use of active antennas, innovative printed circuit board antennas, and other technological advances already in place in existing smartphones, or under development, optimizing current antennas for 600 MHz operations. Duplex Filter: Use of an overlapping duplex filter configuration in the short term until technology allows consolidation into a single filter, likely in the 2015 timeframe Harmonic Interference: Managing the low interference power levels and low likelihood of occurrence of potential harmonic interference into the PCS band by better isolation and harmonic filter technologies, and coordination of 600 MHz and PCS operations. The analysis also reveals that the 35x35 MHz plan shares several technical issues in common with other proposed plans, none of which prevent successful operation or deployment. Importantly, these band plans remove technical issues inherent to the FCC proposed band plan, in particular, the spectral inefficiency resulting from operation of broadcast TV in the duplex gap and the potential for inter-modulation interference caused by high power broadcast TV operations. The characteristics that the 35x35 band plan shares with other similar proposals can be summarized as follows: FINAL Uses 5 MHz paired interchangeable FDD blocks. Uses a 10 MHz Duplex Gap with no TV operation in the Duplex Gap. Allows channel 37 ( MHz) to continue to be used for medical and other unlicensed applications. Provides a total of 10 MHz guard band for broadcast TV channels below 37 (608 MHz). Reinforces the use of 6 MHz guard band for moderate power (50 KW) TV transmission. Reinforces the consensus on re-locating TV operations from the 600 MHz downlink region. Allows for potential use of Supplementary Downlink (SDL) operations below the paired downlink boundary. 44 3/11/2013

157 Suggests relocation of the TV operations from the uplink spectrum because of substantial impact on cellular operations. Allows the use of common antennas for all the operations below 1 GHz, as is currently done for 700 MHz band plans. Exploits many of the mitigation techniques already being used in the 700 MHz band. Suggests extensive coordination with the TV operations in Canada and Mexico to avoid cochannel and adjacent channel interference with US cellular operations in the border areas. The 35x35 MHz band plan proposal focuses on efficient usage of the spectrum, maximum pass band for significant value to the customers and the operators, leverages realistic technological advances, and suggests appropriate protective measures for potential interference both with other LTE systems and TV operations. FINAL 45 3/11/2013

158 6.0 References [3GPP] 3GPP, User Equipment Radio Transmission and Reception, TS a20, [AGIL] Agilent, Filter Design for LTE, Sep pdf?&cc=US&lc=eng [ALU] Alcatel-Lucent comments to FCC on Expanding the Economic and Innovation Opportunities of Spectrum through Incentive Auctions, Docket , Jan 25, 2013 [APT] APT Report, Harmonized Frequency Arrangements for the Band MHz, APT/AWF/REP- 14, Sep 2010, [ATT1] AT&T comments to FCC on Expanding the Economic and Innovation Opportunities of Spectrum through Incentive Auctions, Docket , Jan 25, 2013 [ATT2] AT&T, Intel, NAB, Qualcomm, T-mobile, Verizon letter to FCC on Expanding the Economic and Innovation, Jan 24, 2013 [BINM] Bin-Melha, M, Abd-alhameed R., See, C, Usman, M, Elfergani, R, and Noras, J, Harmonic Rejection Triangular Patch Antenna, Progress in Electromagnetics Research Proceedings, Mar 2012, Piers.org/piersproceedings [CEA] Consumer Electronics Association to FCC on Expanding the Economic and Innovation Opportunities of Spectrum through Incentive Auctions, Docket , Jan 25, 2013 [CEPT] CEPT ECC, Technical Options for the use of a Harmonized Sub-band in the band MHz for Fixed/Mobile Applications (including Uplinks), Dec 21, [CHIU] Chiu, C and Chang, C, A Meandered Loop Antenna for LTE /WWAN Operations in a SmartPhone, Progress in Electromagnetics research C., 2010, [CHO] Cho, I, Lee, I, and Park, Y, Study on Co-existence Between Long Term Evolution and Digital Broadcasting Services, International Journal of Advanced Science and Technology, January, 2012, [CLEA] ClearMobitel, Co-existence of New Services in the 800 MHz band with Digital Terrestrial Television (DTT) A Brief, [CTIA] CTIA to FCC on Expanding the Economic and Innovation Opportunities of Spectrum through Incentive Auctions, Docket , Jan 25, 2013 [EGOR] Egorov, I and Zhinong, Y, A non-uniform helical antenna for dual-band cellular phones, IEEE Antennas and Propagation Society International Symposium, Jul 2000, [EPCO] EPCOS, SAW Duplexer LTE Band 12, Sep [FCC] FCC, Expanding the Economic and Innovation Opportunities of Spectrum Through Incentive Auctions Notice of Proposed Rule Making, Docket No , Oct 2, [GAMP] Gampala, G., Stabler O., Hager, T., Reddy, C, Compact Antenna for LTE Mobile Phone Applications, Microwave Journal, Mar 11, 2012 [GHAR] Gharaibeh K, Non Linear Distortion in Wireless Systems, John Wiles & IEEE, [GOOG] Google Inc. and Microsoft Corp. to FCC on Expanding the Economic and Innovation Opportunities of Spectrum through Incentive Auctions, Docket , Jan 25, 2013 [GUHA] Guha, D. and Antar, Y, Eds., Microstrip and Printed Antennas, Wiley, Feb 2011 FINAL 46 3/11/2013

159 [GUID] Guidotti, A., Barbiroli M., Grazioso, P., Carciofi, C., Guiducci D., and Riva G., Analysis of Coexistence and Mutual Interference Between Mobile and Digital Television Systems, Nov, 2011, [HUFF] Hufford, G, Longley, A, Kissick, W, A guide to the use of the ITS Irregular Terrain Model in the Area Prediction Mode, US Department of Commerce, April 1982, [MAKI] Making Telecoms Work, LTE User Equipment, Network Efficiency, and Value, Sep 2010, [MOBI] MobilMark, Broadband LTE Antennas are Needed for 4G LTE Rollouts, Nov 2011, [MOTO] Motorola to FCC on Expanding the Economic and Innovation Opportunities of Spectrum through Incentive Auctions, Docket , Jan 25, 2013 [NCTA]National Cable and Telecom Association submission to FCC on Expanding the Economic and Innovation Opportunities of Spectrum Through Incentive Auctions, Docket , Jan 25, 2013 [NOKI] Nokia, TV transmission power at UE antenna port, 3GPP TSG-RAN WG4 Meeting # 54, Feb, 2010, [NSN1] Nokia Siemens Network to FCC on Expanding the Economic and Innovation Opportunities of Spectrum through Incentive Auctions, Docket , Jan 25, 2013 [NSN2] 3GPP TSG-RAN WG4 Meeting #54 Contribution R , TV Transmission Power at UE Antenna Port, /ftp/tsg_ran/wg4_radio/tsgr4_54/documents/r zip [QUAL1] Qualcomm submission to FCC on Expanding the Economic and Innovation Opportunities of Spectrum through Incentive Auctions, Docket , Jan 25, 2013 [QUAL2] Qualcomm, Feb 2013, [QUAL3] Reply comments of Qualcomm Inc., Interoperability of Mobile User Equipment Across Paired Commercial Spectrum Blocks in the 700 MHz Band, July 2012, [RAO] Rao, Q and Wang, D, A Compact Dual-Port Diversity Antenna for Long Term Evolution Handheld Devices, IEEE Transactions on Vehicular Technology, Mar 2010, [RNDH] Randhawa B., Parker, I, and Antwi, S, LTE Interference into Domestic Digital Television Systems, COBHAM report, 2010, [SAML] Saml, W, How can Mobile and Broadcasting networks Use Adjacent bands, EBU Technical Review, Mar 2011, [SHMB] Shamblin, J, 4G Devices Demand Active Antenna Solutions, mobile Device Design, Feb 2012, [SKYC] Skycross, SkyCross VersiTune-LTE Antenna Addresses Industry Challenges Brought on by Growing Popularity of 4G LTE, Product Catalogue, [SONG] Song, P. and Barker, D, Antennas, How Many do we Need?, [SPRI] Sprint submission to FCC on Expanding the Economic and Innovation FINAL 47 3/11/2013

160 Opportunities of Spectrum through Incentive Auctions, Docket , Jan 25, 2013 [SUPE] Superconductor Technologies Inc., Eliminate Interference in Your Wireless network while maximizing Your Spectral Efficiency, Coverage, and Data Speeds, 2013, [TMOB1] T-Mobile submission to FCC on Expanding the Economic and Innovation Opportunities of Spectrum through Incentive Auctions, Docket , Jan 25, 2013 [TMOB2] T-Mobile FCC review package 1/30/13 [VERI] Verizon submission to FCC on Expanding the Economic and Innovation Opportunities of Spectrum through Incentive Auctions, Docket , Jan 25, 2013 [WONG] Wong, K and Chen, W, Small sized Printed Loop-Type Antenna Integrated with two stacked Coupled-Fed Shorted Strip Monopoles for eight Band LTE/GSM/UMTS Operation in the mobile, Microwave and Optical Technology Letters, Jul 2010, FINAL 48 3/11/2013

161 Appendix A: Other Band Plan Proposals This section contains descriptions of other band plans that have been proposed and evaluated for application to the 600 MHz incentive auction spectrum. These descriptions are provided to facilitate comparisons to the 35x35 MHz plan made in Sections 3-6. A.1 25x25 MHz Plan A 25x25 MHz band plan has been proposed in [ATT1] and [QUAL1]. In the initial band plan proposed in [ATT1], a duplex gap of 14 MHz is used with spectrum, providing for two, 6 MHz TV channels above Channel 37 (see Figure A.1). Figure A.1: A Representative 25x25 MHz Band Plan FINAL 49 3/11/2013

162 A.2 30x30 MHz Plan A 30x30 MHz band plan which allows for one TV station above channel 37 has also been proposed and considered by [QUAL1], [ALU],.(See Figure A.2) Figure A.2: A Representative 30x30 MHz Band Plan In both the 25x25 MHz and 30x30 MHz plans, TV operation with its associated 8 MHz guard Band (GB) is not the only option for using the spectrum above channel 37 (614 MHz). It is envisioned that if sufficient broadcast TV channels can be cleared, the spectrum between 614 MHz and the lower end of the proposed downlink), which is 634 MHz and 628 MHz for the 35x35 MHz and the 30x30 MHz band plans respectively, can be used for other applications, for example, supplementary downlink (SDL), or secondary unlicensed applications. FINAL 50 3/11/2013

163 Appendix B: Company Profile Profile: Roberson and Associates, LLC Roberson and Associates, LLC, is a technology and management consulting company serving government and commercial customers that provides services in the areas of RF spectrum management, RF measurements and analysis, strategy development, and technology management. The organization was founded in 2008 and is composed of a select group of individuals with corporate and academic backgrounds from Motorola, Bell Labs (AT&T, Telcordia, Lucent and Alcatel-Lucent), IBM, IITRI (now Alion), independent consulting firms, and Illinois Institute of Technology. Together the organization has over 300 years of the high technology management and technical leadership experience with a strong telecommunications focus. Profiles: Roberson and Associates, LLC, Staff Dennis A. Roberson, President and CEO, Roberson and Associates Mr. Roberson is the Founder, President and CEO of Roberson and Associates, LLC. In parallel with this role he serves as Vice Provost and Research Professor in Computer Science at Illinois Institute of Technology where he has responsibility for IIT s corporate relationships including IIT s Career Management and Technology Transfer efforts. He also supports the implementation of IIT s Strategic Plan, the development of new research centers, and the successful initiation and growth of IIT related technology-based business ventures. He is an active researcher in the wireless networking arena and is a co- founder of IIT s Wireless Network and Communications Research Center (WiNCom). His specific research focus areas include dynamic spectrum access networks, spectrum occupancy measurement and spectrum management, and wireless interference and its mitigation and of which are important to the Roberson and Associates mission. He currently serves on the governing and / or advisory boards of several technology-based companies. Prior to IIT, he was EVP and CTO at Motorola and he had an extensive corporate career including major business and technology responsibilities at IBM, DEC (now part of HP), AT&T, and NCR. He is and has been involved with a wide variety of Technology, Cultural, Educational and Youth organizations currently including the FCC Technical Advisory Council and Open Internet Advisory Committee, the Commerce Spectrum Advisory Committee, and the National Advisory Board for the Boy Scouts of America and its Information Delivery Committee, and the Board of HCJB Global. He is a frequent speaker at universities, companies, technical workshops, and conferences around the globe. Mr. Roberson has BS degrees in Electrical Engineering and in Physics from Washington State University and a MSEE degree from Stanford. Kenneth J. Zdunek, Ph.D., V.P. and Chief Technology Officer Dr. Zdunek is Vice President and the Chief Technology Officer of Roberson and Associates. He has 35 years of experience in wireless communications and public safety systems. Concurrently he is a research faculty member in Electrical Engineering at the Illinois Institute of Technology, in Chicago, Illinois, where he conducts research in the area of dynamic spectrum access and efficient spectrum utilization, and teaches a graduate course in wireless communication system design. He is a Fellow of the IEEE, recognized for his leadership in integrating voice and data in wireless networks. Prior FINAL 51 3/11/2013

164 to joining Roberson and Associates, he was VP of Networks Research at Motorola, a position he held for 9 years. Dr. Zdunek was awarded Motorola s patent of the year award in 2002 for a voice-data integration approach that is licensed and extensively used in GSM GPRS. He holds 17 other patents, included patents used in public safety trunked systems and cellular and trunked systems roaming. He directed the invention and validation of Nextel s iden TM voice-data air interface and IP based roaming approach, and was the principal architect of Motorola s SmartNet TM public safety trunking protocol suite. In the 1990 s, he directed a Spectrum Utilization and Public Safety Spectrum Needs Projection submitted to the FCC in support of the 700 MHz spectrum allocation for Public Safety. He was awarded the BSEE and MSEE degrees from Northwestern University, and the Ph.D. EE degree from the Illinois Institute of Technology. He is a registered Professional Engineer in the State of Illinois. Suresh R. Borkar, Ph.D. Senior Principal Investigator Dr. Borkar is a Senior Principal Investigator at Roberson and Associates and a member of the faculty in the Electrical and Computer Engineering (ECE) department at the Illinois Institute of Technology (IIT), Chicago. Previously, he was with AT&T/Lucent Technologies/Alcatel-Lucent (ALU) for over 26 years responsible for various facets of product management, systems engineering, architecture, development, integration and testing, and customer management in Computer and Networking systems, Wireline Switching systems, Data systems, and Wireless systems. He was the Director for Customer Management for 3G mobility systems responsible for customer positioning, acceptance, and revenue realization. He was previously the Chief Technology Officer (CTO) and Managing Director, Lucent India Inc., responsible for all Lucent customer products and business activities in India. Dr. Borkar develops knowledge share and teaches advanced courses in Telecommunications and Computer Architecture for the Academia, IEEE, and the industry. He has been an organizer and moderator of conferences and panel discussions on WiMAX and VoIP/Next Generation Networks (NGNs). Dr. Borkar received his B. Tech. in Electrical Engineering from Indian Institute of Technology Delhi (India) and M.S. and Ph. D. in ECE from Illinois Institute of Technology, Chicago. FINAL 52 3/11/2013

165 Exhibit B

166 Spectrum Auction Rules That Foster Mobile Wireless Competition Jonathan B. Baker March 12, 2013 In the Matter of Policies Regarding Mobile Spectrum Holdings, WT Docket No I. Introduction A. Qualifications I am a Professor of Law at American University s Washington College of Law. In 2011, I served as Senior Economist for Transactions at the Federal Communications Commission ( Commission ), and prior to that I was the Commission s Chief Economist for approximately two years. From 1995 to 1998, I served as the Director of the Bureau of Economics at the Federal Trade Commission. I have also worked as a Senior Economist at the President s Council of Economic Advisers, Special Assistant to the Deputy Assistant Attorney General for Economics in the Antitrust Division of the Department of Justice, an Assistant Professor at Dartmouth s Amos Tuck School of Business Administration, an Attorney Advisor to the Acting Chairman of the Federal Trade Commission, and an antitrust lawyer in private practice. I am co-author of an antitrust casebook, a past Editorial Chair of the Antitrust Law Journal, and a past member of the Council of the American Bar Association s Section of Antitrust Law. I have published widely in the fields of antitrust law and policy and industrial organization economics. In 2004, I received American University s Faculty Award for Outstanding Scholarship, Research, and Other Professional Accomplishments, and in 1998 I 1 T-Mobile is also submitting this paper in WT Docket No , In the Matter of Expanding the Economic and Innovation Opportunities of Spectrum Through Incentive Auctions, in response to the Commission s request for comments on how to structure the forward auction of spectrum reclaimed from broadcasters.

167 received the Federal Trade Commission s Award for Distinguished Service. I hold a J.D. from Harvard and a Ph.D. in Economics from Stanford University. B. Assignment The Commission is seeking comment on whether and how to revise its rules and policies applicable to the acquisition of spectrum by mobile wireless carriers. 2 Among other things, the NPRM requests comment on whether to use a case-by-case analysis for spectrum acquisitions or to prefer bright-line limits; whether to include additional spectrum bands in evaluating spectrum holdings; how to analyze geographic markets; and whether the Commission should make distinctions among spectrum bands in assessing spectrum holdings. 3 I have been asked by T-Mobile to review the Commission s NPRM and comments filed to date by various parties, and discuss relevant economic considerations for evaluating spectrum holdings. In particular, I have been requested to provide an economic analysis of T-Mobile s proposal for spectrum caps in auctions of new spectrum and case-by-case reviews for secondary market transactions, and to contrast it with an approach in which every transaction is reviewed individually. I have also been asked to discuss some conceptual issues related to the treatment of different spectrum bands. My submission will not provide a comprehensive analysis of all issues raised by the NPRM. C. Main themes My main conclusions are the following: i. Rules restricting spectrum aggregation at the time of new spectrum auctions can foster competition in services that use wireless spectrum as an input. 2 See Policies Regarding Mobile Spectrum Holdings, Notice of Proposed Rulemaking, 27 FCC Rcd (2012) ( NPRM ). The NPRM emphasizes the need for rules of the road that are clear and predictable that promote the competition needed to ensure a vibrant, world-leading, innovation-based mobile economy. Id Id.; see also Expanding the Economic and Innovation Opportunities of Spectrum Through Incentive Auctions, Notice of Proposed Rulemaking, 27 FCC Rcd (2012). 2

168 ii. iii. iv. Spectrum caps are likely to avoid the costs and delays that would result from the use of case-by-case reviews of spectrum acquisitions after auctions, and are also likely to prevent efficiency-reducing distortions in spectrum allocation arising from uncertainty about the outcome of postauction review. Spectrum caps in auctions can encourage auction participation, so they have the potential to increase auction revenues. Case-by-case reviews are more troublesome when applied to auction outcomes than when applied later to secondary market transactions in spectrum. v. Case-by-case reviews of secondary market transactions would be unlikely to encourage speculative bidding that could otherwise undermine spectrum caps in auctions. vi. Separate caps for low-frequency spectrum can be beneficial even if wireless providers can overcome disadvantages of high-frequency spectrum with sufficient capital investment. II. Economic Considerations in Developing Spectrum Auction Rules A. Fostering wireless competition When spectrum ownership is concentrated, firms may be able to exercise market power downstream in the provision of services that use wireless spectrum as an input. Large incumbent firms that recognize this prospect may have an incentive and ability to obtain or maintain downstream market power by keeping spectrum away from their rivals. 4 When spectrum is auctioned, the foreclosure value that large incumbents may place on spectrum acquisitions can distort spectrum allocations and downstream competition. If the incumbent can limit competition from excluded rivals by acquiring a spectrum block at auction, the value it will place on that spectrum will include its market power benefit, and will therefore 4 In general, this incentive and ability would be expected to increase with a firm s market share and with aggregate market concentration. 3

169 exceed the social value of the spectrum acquisition. 5 In consequence, these firms may outbid rivals and succeed in obtaining or maintaining market power in downstream services, when that would not be the best outcome for consumers or society as a whole. Spectrum policies, such as auction rules that incorporate spectrum ownership caps, can limit or prevent such competitive distortions. Moreover, spectrum policies that would address this problem do not necessarily sacrifice substantial economic benefits, even if a firm s greater scale within a market confers production efficiencies. If the foreclosed rivals are limited in their ability to achieve scale economies, that will limit the investments they make and the competitive constraint they will impose on the large incumbents, and thus limit the extent to which any benefits of increased scale to large incumbents are passed on to consumers in the form of lower prices, higher quality service, or new service offerings. 6 5 Peter Cramton, Evan Kwerel, Gregory Rosston & Andrzej Skrzypacz, Using Spectrum Auctions to Enhance Competition in Wireless Services, 54 J.L. & ECON. 167, (2011) ( [A]n auction that awards the spectrum to bidders with the highest values may not assure economic efficiency because the bidders private values for the spectrum may differ from social values as a result of market structure issues. For example, an incumbent will include in its private value not only its use value of the spectrum but also the value of keeping the spectrum from a competitor. ) (internal citations omitted); Ex Parte Presentation of United States Department of Justice, GN Docket No , at (filed Jan. 4, 2010) ( The goal in assigning licenses to any such new spectrum designated for commercial services should be to ensure that it generates the greatest ultimate benefits to the consumers of those services. When market power is not an issue, the best way to pursue this goal in allocating new resources is typically to auction them off, on the theory that the highest bidder, i.e., the one with the highest private value, will also generate the greatest benefits to consumers. But that approach can go wrong in the presence of strong wireline or wireless incumbents, since the private value for incumbents in a given locale includes not only the revenue from use of the spectrum but also any benefits gained by preventing rivals from eroding the incumbents existing businesses. The latter might be called foreclosure value as distinct from use value. The total private value of spectrum to any given provider is the sum of these two types of value. However, the foreclosure value does not reflect consumer value; to the contrary, it represents the private value of forestalling entry that threatens to inject additional competition into the market. In an established oligopoly with large margins between the price and the incremental cost of existing broadband services, the foreclosure value for incumbents in a given locale could be very high. ) 6 See generally Jonathan B. Baker, Beyond Schumpeter vs. Arrow: How Antitrust Fosters Innovation, 74 ANTITRUST L.J. 575 (2007) (discussing the benefits of competition for innovation). 4

170 B. Taking a long-term perspective Communications markets have changed dramatically since spectrum was first allocated early in the 20th century. New technologies and products have changed how communications services are provided and what buyers demand, leading to shifts in the nature of services that provide the most valuable use of various ranges of spectrum. Spectrum policies should take a long-term perspective: they should recognize that communications markets are likely to continue to change rapidly, so the best uses of spectrum today may not be the best in the future and the best future uses may not be apparent today. 7 Changing spectrum uses are likely to exacerbate long-run problems associated with excessive spectrum aggregation. If a small number of incumbent providers end up with control over large amounts of spectrum, those incumbents may have the incentive and ability to frustrate the development of new technologies and business models brought to the market by smaller rivals and potential competitors (including future rivals that cannot now be identified), thereby preventing or delaying the development of new competition. The resulting competitive harms may not be limited to downstream markets in which producers use spectrum as an input; they may also extend to markets in complementary products and services (e.g., wireless infrastructure and device vendors, wholesale wireless services, and mobile applications). Policies to ensure greater long-run competition in wireless services may benefit complementary markets, as by increasing the demand for complementary products those that exist today and those that will be developed in the future. 8 With future technology and demand uncertain, auction rules and limitations on secondary market transactions can be an important tool for protecting long-term competition in markets for 7 Cf. Evan Kwerel & John Williams, Changing Channels: Voluntary Reallocation of UHF Television Spectrum (FCC Office of Plans and Policy Working Paper 27, Nov. 1992) (documenting the potential welfare gains of shifting spectrum from broadcast television services to mobile wireless services in the early 1990s). 8 Furthermore, greater competition in downstream wireless markets will likely contribute to achieving the large economic benefits of mobile wireless services, including mobile broadband services. See COUNCIL OF ECON. ADVISERS, EXEC. OFFICE OF THE PRESIDENT, THE ECONOMIC BENEFITS OF NEW SPECTRUM FOR WIRELESS BROADBAND, at (2012), available at (growth in mobile broadband is likely to generate substantial economic benefits including GDP growth, job growth, and productivity gains). 5

171 services that use spectrum resources, as well as competition in complementary markets. The alternatives for achieving these long term goals are unattractive: it is commonly impractical to reallocate spectrum by regulatory fiat, 9 and spectrum owners that can exercise market power in downstream services as a result of excessive spectrum aggregation cannot be expected to give up that market power through voluntary spectrum transactions in secondary markets. When developing policies to limit spectrum concentration in order to prevent long-run competitive harms in an environment dominated by uncertainty about future technologies and spectrum uses, the Commission would almost necessarily employ similar standards (such as a maximum ownership percentage for various spectrum bands in a market) to review auction outcomes, regardless of whether the Commission promulgates those standards by rule in advance or employs them as the basis of case-by-case reviews of auction allocations after the auction has taken place. 10 Were the Commission instead to make long-run competitive judgments based on market characteristics that depend on current spectrum uses, it would risk frustrating the development of new technologies or business models. Accordingly, there may be little or no advantage in relying on post-auction review (relative to incorporating spectrum aggregation standards in auction rules) to compensate for the greater distortions, inefficiencies and transaction costs discussed in Sections III and IV below. 9 Spectrum licenses are generally renewed without complication. This approach helps ensure that licensees have an incentive to make investments in providing service that might not pay off during the license period. Moreover, as a general rule, spectrum reallocations by fiat are likely less efficient than secondary market transactions in shifting spectrum to its best use, if secondary market participants are discouraged from achieving or preserving market power through those transactions. 10 This is neither a claim about the relative advantages of framing spectrum ownership policies in the form of bright-line rules versus unstructured standards, nor a claim about the relative advantages of establishing prophylactic policies versus waiting until uncertainty about the future is clarified. The point is simply that structural factors tied to current technologies and spectrum uses, perhaps including the growth rates and excess capacity of market participants at the time of the spectrum transfer, however relevant to assessing short-run competitive dynamics, have limited predictive value for assuring long-run competition when future technologies and spectrum uses are uncertain. In consequence, the Commission would likely frame policies for fostering long-run competition primarily on factors that are not tied to current spectrum uses, such as ownership shares of various types of spectrum, regardless of whether those policies are implemented through auction rules or through post-auction caseby-case review. Doing so would not limit the Commission s flexibility to respond to changing circumstances. For example, the spectrum caps employed for a particular auction could be revised over time to adapt to changing circumstances, as the Commission balances the benefits of stable regulatory policy for encouraging investment against the benefits of adjustment to new information and adaptation to changed circumstances. 6

172 C. Generating auction revenues Although my report is mainly concerned with how auction policies could achieve wireless competition goals, I will also comment briefly on their implications for auction revenues. Most importantly, there is no necessary tradeoff between fostering wireless competition through auction restrictions and generating high auction revenues. 11 Prices paid by the winning bidders depend on many factors including the type and quality of spectrum auctioned, the structure of the auction, the number of bidders, the valuation that bidders place on the spectrum offered, and bidders budget constraints. Auction restrictions can affect many of these factors. 12 While spectrum caps could reduce the quantity of spectrum that some bidders would demand, restrictions on the ability of large firms to bid could increase auction participation and the overall quantity demanded, potentially increasing auction revenues. 13 I will further discuss the possibility that spectrum caps could increase auction participation in the section below. III. Economic Analysis of T-Mobile s Suggested Approach A. Benefits of T-Mobile s approach T-Mobile recommends that the Commission address the problem of excess spectrum agglomeration through spectrum caps when auctioning new spectrum, 14 and through case-bycase reviews of secondary market transactions involving spectrum. This proposal sensibly reflects differences between the two settings that call for a different tradeoff between precision and certainty. 11 See Paul Klemperer, How (Not) to Run Auctions: the European 3G Telecom Auctions, 46 EUR. ECON. REV. 829 (2002) (survey of European spectrum auctions indicates absence of a general relationship between auction restrictions and the revenues generated). 12 See Ian Ayres & Peter Cramton, Deficit Reduction Through Diversity: How Affirmative Action at the FCC Increased Auction Competition, 48 STAN. L. REV. 761 (1996). 13 The consequences of spectrum caps for revenues in any particular auction setting would depend on which effect is the more important influence in that setting. 14 A prohibition on warehousing spectrum would not substitute for spectrum caps, because rivals could be foreclosed and competition harmed even if the licensee used its spectrum to provide service. 7

173 In an auction setting, a spectrum cap is preferable to case-by-case reviews of auction outcomes, as it provides clear guidance to firms bidding in auctions and limits the transaction costs of regulation. 15 Auctions often involve many parties, each making interdependent decisions regarding multiple alternatives available for sale at the same time. This complexity amplifies the importance of clear auction rules with certain application for achieving efficient spectrum allocation. Absent clear auction rules, firms may base their bids on potentially erroneous predictions of how the agency will react in an after-the-fact review of auction results, distorting auction bidding and outcomes. 16 Once the auction is over, auction winners and losers often would make commitments to business plans that would change the valuation of any spectrum that might be reallocated in a post-auction review. Hence, the likely remedies if the Commission concludes that it should not permit an auction winner to acquire the spectrum block on which it submitted the highest bid a rerun of the entire auction or a required divestiture in the secondary market would likely lead to a different outcome than would have been obtained had the disqualified firm been prevented from bidding in the first place by a spectrum cap. If a full auction rerun is impractical, moreover, and the firm required to divest spectrum is permitted to choose which bands to divest or select the new owner, it would be able to make those choices in ways that reduce potential competition to itself, further enhancing the inefficiency of the resulting spectrum allocation. By laying out clear rules governing the initial auction, the Commission would prevent such distortions. Auction rules also avoid the cost and time involved with regulatory reviews after the 15 It would be wrong to picture spectrum caps as rigid screens that would need later correction through caseby-case reviews. Spectrum caps can be fine-tuned auction-by-auction to address changing conditions, such as increases in suitable and available spectrum. 16 Erroneous predictions could be minimized if the Commission applies clear, detailed, and well-specified criteria to review auction results after the auction is completed. But if the Commission does that, there would be no advantage in applying those criteria in an after-auction review rather than in advance. To the extent the criteria are instead flexible in interpretation or ambiguous, erroneous predictions and the resulting distortions in spectrum allocation would arise notwithstanding the best efforts of bidding firms and their outside counsel to forecast the outcome of after-action reviews. In an analogous context the efforts merging firms and their outside counsel make to determine how the antitrust agencies will treat proposed acquisitions the evidence shows that outside parties can be surprised by policy changes for several years before catching on fully. See Jonathan B. Baker & Carl Shapiro, Detecting and Reversing the Decline in Horizontal Merger Enforcement, 22 Antitrust A.B.A. 29, 31 (2008) (using merger enforcement statistics to identify two four-year periods when the Antitrust Division surprised the antitrust bar with their lack of interest in challenging mergers). 8

174 auction has taken place, as well as any additional distortions associated with prolonging the uncertainty about how spectrum would be allocated. Case-by-case reviews to avoid excessive spectrum agglomeration make more sense for secondary market transactions, regardless of whether they take place as a two-party negotiation or an informal auction. Secondary transactions permit the reallocation of spectrum to more efficient uses as time passes and circumstances change since the original spectrum allocation. At that later time, delays in resolving the ownership of the spectrum would be less costly, and lack of certainty less troublesome, than with reviews of spectrum allocations resulting from large scale multi-player auctions. The additional precision allowed by an individualized review of secondary transactions may outweigh the costs of such approach, even though the cost-benefit calculus would differ in the auction setting. B. Critiques of approach Some commenters have claimed that T-Mobile s recommended approach of treating initial auctions differently from secondary transactions is unworkable and would allow arbitrage profiteering. 17 Their main theory, as I understand it, is that the spectrum caps would be circumvented by smaller firms not bound by the cap. Those firms would make speculative purchases of spectrum in initial auctions with the main purpose of flipping it to firms for which the cap is binding. In this story, the small firms would reap profits at the expense of government revenues. I find this argument unconvincing. The flipping argument supposes that a bidder that could not bid in the auction due to the cap would prevail in a case-by-case review of its purchase of the same spectrum in a secondary market transaction in the wake of the auction. This is unlikely because the Commission would be expected to apply similar principles to case-by-case reviews that it would apply in determining the initial spectrum cap (as I discuss further in Section IV.C). It is particularly difficult to imagine the Commission applying different standards when conducting a case-bycase review not long after a spectrum auction, the time when a pure speculative bidder would 17 E.g. Reply Declaration of Mark A. Israel and Michael L. Katz, WT Docket No , at (Jan. 7, 2013), attached as Attachment B to Reply Comments of AT&T, Inc., WT Docket No (filed Jan. 7, 2012) ( Israel and Katz Reply Declaration ). 9

175 find it most profitable to sell. Moreover, speculative bidders, interested in spectrum only for the purpose of resale, would be unlikely to accept the significant risk that the Commission would not approve a secondary transaction with a firm that was prohibited from bidding in the initial auction, particularly when speculators also recognize that they must bear the costs of participating in the auction, the costs of negotiating a resale, the costs of participating in a Commission proceeding reviewing the transfer of the spectrum to a buyer unable to purchase it initially, and the cost arising from the time it takes to negotiate resale and resolve the Commission s review. 18 These risks and the transaction costs would likely deter speculative bidding. 19 C. Impact on auction participation and revenues T-Mobile s suggested approach has the additional benefit of encouraging auction participation, which I understand to be one of the Commission s goals. A firm that must bid against a large incumbent that would obtain a foreclosure value from acquiring the spectrum may expect to be outbid for that reason. Given the non-trivial fixed costs of auction participation, a firm expecting to be outbid could readily be deterred from participating in the auction in the first place. If auction participation is thin as a result of this dynamic, the large incumbent firms that are in principle willing to pay to obtain foreclosure benefits may enjoy these benefits without bidding up the auction price to a level that pays for those benefits fully, leaving the public with a less competitive wireless sector and the government with lower revenues than could be obtained. Under such circumstances, auction rules, such as spectrum caps that curb bidding for the sake of foreclosure, would encourage entry into the auction by potentially-foreclosed rivals. 18 Delay would be costly for a speculative purchaser. It would bear either a financing cost or an opportunity cost on the capital it has invested. Moreover, if it is required to offer service by Commission auction rules, it may be required to make substantial irreversible investments if it does not flip the spectrum quickly. 19 Even if spectrum flipping were to occur, moreover, it may not shift significant revenues from the government to spectrum speculators. In this hypothetical scenario, there would be many speculative bidders, and speculators would bid more aggressively in the initial auction when they see a potential for receiving a high value in the secondary market under a flipping scenario. The resulting bidding competition would be expected to limit the windfall that speculators would receive to a discount reflecting their participation costs and risks. 10

176 Since auction revenues generally increase with auction participation, the increase in revenues associated with enhanced participation could offset, or more than offset, the revenue effect of the reduced spectrum demand from large incumbents with holdings that exceed the cap. This mechanism, by which spectrum caps may increase auction revenues, is well understood in the auction economics literature, both on theoretical grounds and as demonstrated empirically. 20 Spectrum caps also have an advantage over case-by-case post-auction reviews in raising auction revenues by encouraging more aggressive bidding by large incumbent firms that would be under the cap in various markets but would be uncertain about the outcome of a post-auction review. With post-auction review, those firms would discount their bids to account for the risk that they might later bear the costs of divesting the spectrum they have won. This possibility could be prevented by rules applied to assess bidding eligibility in advance of the auction, such as spectrum caps. IV.The Risks of Case-by-Case Reviews in Auctions This section discusses in more detail several problems that would arise if the Commission chose to address the problem of excessive spectrum agglomeration through post-auction reviews rather than through spectrum caps. A. False positives and false negatives Section III.A of this report explained that when bidders lack clear guidance about whether they will be barred from acquiring spectrum blocks, their uncertainty about the results of an 20 Peter Cramton, Spectrum Auctions, in HANDBOOK OF TELECOMMUNICATIONS ECONOMICS 605, 631 (Martin Cave, Sumit Majumdar & Ingo Vogelsang eds., 2002) ( Typically, spectrum caps lower auction revenues, but there is one important exception. In situations where incumbent bidders have an advantage, a spectrum cap may actually increase revenues and promote efficiency. In such a situation without a spectrum cap, non-incumbents may be unwilling to participate in the auction, knowing that the incumbents will ultimately win. As a result, in the auction without the cap only the incumbents show up, there is a lack of competition, and the incumbents split the licenses up among themselves at low prices. With the cap, the non-incumbents know that non-incumbents will win licenses, giving them the incentive and ability to secure the needed financing from capital markets. A competitive auction with market prices results. This phenomenon of incumbent bidders getting good deals, because of a lack of nonincumbent competition has been seen in some US auctions, but is most vivid in the Dutch UMTS auction. ). See also Cramton, Kwerel, Rosston and Skrzypacz, supra note 5, at 174 (2011) ( revenues in unrestricted auctions do not need to be strictly higher than those in auctions with spectrum caps or set-asides ). 11

177 after-the-fact review could distort their bidding and lead to an inefficient allocation of spectrum. A case-by-case approach may lead to false positives and false negatives. A false positive arises if a firm that wins in an auction is prohibited from retaining the spectrum it acquired after a regulatory review. The costs associated with such an outcome are potentially substantial, as rivals that might have won in the initial auction but were outbid by the large firm (or decided not to bid because they expected to be outbid) may have found workarounds by the time the winning firm is required to divest. These firms may not bid for the divested spectrum, and the eventual acquirer of the spectrum may be different from the bidder that would have obtained the spectrum had the initial auction not been distorted. The latter concern would be exacerbated by the ability of the divesting firm to select the firm that would purchase the divested spectrum. Such a situation would result in an inefficient spectrum allocation and wasteful transaction costs. A false negative arises if a firm that would have bid and won does not bid because it falsely believes that its acquisition may be rejected, or fails to win because it reduces its bid due to the risk of post-auction divestitures. In either case, this will result in an inefficient spectrum allocation and likely lower auction revenues. 21 B. Time-inconsistency distortions While the spectrum misallocation situations described above are perhaps the most transparent potential distortions that could result from relying on a case-by-case review to address excessive spectrum agglomeration, they are not the only or even the primary concerns arising from post-auction review. Distortions created by case-by-case reviews could extend to situations in which the auction result is not reversed in a subsequent review and firms were not deterred from participating. In particular, outcomes may be distorted if firms bid even though their acquisition of spectrum would likely harm wireless competition, on the hope that they could later convince the Commission to approve their acquisition. Under a case-by-case review system, by the time the 21 If the auction rules include a safe harbor for firms that might otherwise be uncertain about the outcome of a post-auction review, however, this will reduce or eliminate the likelihood of false negatives. 12

178 Commission has to make its decision about approving a spectrum purchase, one set of distortions the distortions in the outcome of the auction itself has already been created. Moreover, reversing the auction could result in significant delays in the spectrum deployment, and to complaints about the penalty the acquiring firm must pay if it resells the licenses at a loss. A forward-looking Commission might consider these distortions as sunk and approve an acquisition that it would have earlier considered to be anti-competitive. Firms that foresee this possibility could take advantage of the Commission s time-inconsistency by bidding for spectrum that they would be prohibited from acquiring by a spectrum cap, knowing that their anticompetitive purchases will be too costly to reverse. Under such circumstances, post-auction reviews would impose costs on firms without adequately protecting competition. C. Spectrum cap v. case-by-case review approach If the Commission would frequently reach a different and better outcome through postauction case-by-case reviews compared to the outcomes it would reach by specifying a spectrum cap as part of its auction rule, then a case-by-case approach would warrant closer consideration. This situation is unlikely often to occur, however. A case-by-case review could not practically avoid applying general guidelines for preventing undue spectrum concentration, and those guidelines are unlikely to differ markedly from those that would be specified in developing a spectrum cap. By using a spectrum cap, the Commission can implement those guidelines while avoiding the inefficiencies and distortions associated with post-auction review described above. 22 Accordingly, any potential advantage of case-by-case review of auction outcomes in the precision with which it can identify harmful spectrum concentration is likely limited, and outweighed by the disadvantages associated with the distortions inherit in this approach. A caseby-case approach for auctions is unlikely to do better in protecting long-run competition than a spectrum cap and is likely to prove more costly to implement when compared with spectrum caps. For that reason, T-Mobile s suggested approach for fostering competition in mobile 22 As previously discussed, supra note 15, spectrum caps can be fine-tuned to address changing conditions, or even waived in unique circumstances, so are unlikely to perform substantially worse than case-by-case reviews in targeting competitive concerns. 13

179 wireless services is likely to perform better than an across-the-board case-by-case approach under most circumstances. V. Should All Bands of Spectrum Be Treated Equally? I have not conducted an in-depth study of the relative merits of different bands of spectrum, but it is widely understood that different frequency bands have different attributes, and in particular that spectrum below 1 GHz possesses particularly valuable properties for mobile wireless services providers. 23 In this section, I will share some observations about the benefits of imposing separate spectrum caps for different spectrum bands on account of these differences. A. Complementarities between bands of spectrum The different characteristics of low-frequency spectrum bands and high-frequency spectrum bands make them complements in providing mobile wireless services. In general, wireless providers think of lower frequency spectrum as better suited for expanding a wireless network s coverage (because a base station offering service in a lower frequency band will have greater geographic coverage and superior in-building penetration than a similar station offering service in a higher band), and higher frequency spectrum as better suited for expanding a network s capacity (as demand grows for existing services, for example). In consequence, mobile wireless services of any given geographic coverage and quality and typically can be 23 Implementation of Section 6002(b) of the Omnibus Budget Reconciliation Act of 1993; Annual Report and Analysis of Competitive Market Conditions With Respect to Mobile Wireless, Including Commercial Mobile Services, Fifteenth Report 26 FCC Rcd (2011) ( Fifteenth Competition Report ) (noting that lower frequency bands (below 1 GHz) possess more favorable intrinsic spectrum propagation characteristics than spectrum in higher bands, allowing mobile wireless providers using those band to provide superior coverage over larger geographic areas, through adverse climates and terrain, and inside buildings and vehicles ); Application of AT&T Inc. and Qualcomm Incorporated For Consent To Assign Licenses And Authorizations, Order, 26 FCC Rcd (2011) ( AT&T and Qualcomm Order ) (explaining that spectrum resources in different frequency bands can have widely disparate technical characteristics that affect how the bands can be used to deliver mobile services and noting that the propagation characteristics of spectrum below 1GHz, as allow for better coverage across larger geographic areas and inside buildings while the higher frequency spectrum is ideal for delivering advanced wireless services to rural areas ). 14

180 provided more efficiently using a mix of low and high spectrum frequencies than using either frequency exclusively. 24 The cost penalty for providing service without using a mix of spectrum frequencies is not symmetric: it is likely to be particularly high for providers that mainly employ high-frequency spectrum, with limited use of low-frequency spectrum. 25 Low-frequency spectrum can serve the capacity function more typically associated with high-frequency spectrum. But the physical properties of high-frequency spectrum make it costly and less practical for wireless providers to use high-frequency spectrum to serve the coverage function more typically associated with lowfrequency spectrum. Under such circumstances, a wireless provider may disadvantage rivals (raising their production costs) by denying them access to low-frequency spectrum, even if highfrequency spectrum can physically substitute for low-frequency spectrum to some extent with additional capital investment. Accordingly, the Commission should consider whether excessive agglomeration of low-frequency spectrum in the hands of large incumbents would constrain the ability of those rivals with limited access to low-frequency spectrum to compete aggressively in wireless services markets, and thereby allow the large incumbents to obtain or maintain market power, independent of its concern about the competitive consequences of excessive aggregation of wireless spectrum overall. 24 See, e.g., Fifteenth Competition Report 307 ( [G]iven the superior propagation characteristics of spectrum under 1 GHz, particularly for providing coverage in rural areas and for penetrating buildings, providers whose spectrum assets include a greater amount of spectrum below 1 GHz spectrum may possess certain competitive advantages for providing robust coverage when compared to licensees whose portfolio is exclusively or primarily comprised of higher frequency spectrum. As discussed above, holding a mix of frequency ranges may be optimal from the perspective of providing the greatest service quality at low cost. ); Implementation of Section 6002(b) of the Omnibus Budget Reconciliation Act of 1993, Fourteenth Report, 25 FCC Rcd (2010) ( Fourteenth Competition Report ). 25 See, e.g., AT&T and Qualcomm Order 49 (2011) ( AT&T itself has recognized this distinction [between low- and high-frequency spectrum] in the context of its bid to acquire T-Mobile, where it asserted that a significant benefit to T-Mobile customers would be their newly acquired access to AT&T spectrum below 1 GHz, enabling those customers to receive both extended rural coverage and superior in-building and in-home service due to access to AT&T s spectrum below 1 GHz. ). 15

181 B. Trade-off between spectrum attributes and capital investments Some commenters in this proceeding have suggested that competition would not be affected if some firms are limited to high-frequency spectrum holdings, as those firms would be able to acquire their spectrum assets at a lower cost and then apply these cost savings towards greater infrastructure investments. By doing this, they say, the wireless provider would duplicate the scope and quality of service offered by a firm that has substantial holdings of low-frequency spectrum for a cost that is similar overall. Therefore, their argument goes, differences in attributes of spectrum bands are unimportant, and there is no justification for separate restrictions on each band in order to protect competition. I disagree. Even if it were true that differences in spectrum prices exactly offset the difference in discounted present value of the capital investments needed to equalize the quality of service (a proposition I have not evaluated empirically), large incumbents could still obtain or maintain market power by foreclosing rivals from access to low-frequency spectrum, and separate spectrum caps by band could still benefit competition more effectively than a single cap on overall spectrum. This possibility arises because firms need to decide not only whether to build out using their spectrum, but also how much to spend on doing so. 26 To see this point, suppose that in a hypothetical competitive auction a setting that rules out the possibility that large incumbents would obtain or maintain market power by foreclosing their rivals a firm would choose to purchase low-frequency spectrum and use that spectrum to offer service comparable in quality and scope to what the large incumbents provide (which I will term equivalent service). High-frequency spectrum is also available, at a lower price, but at that price differential, the firm would prefer to purchase and build out low-frequency spectrum. Now relax the assumption that large incumbent could not take into account the foreclosure value of bidding for the low-frequency spectrum and suppose it bids up the price of low-frequency spectrum to the point where the rival firm prefers instead to purchase high- 26 The hypothetical example suggested by Israel and Katz involving a backyard swing set that sells at a lower price disassembled misleads because it does not allow for this possibility. Israel and Katz Reply Declaration, supra note 17, at In their example, the family buying a disassembled swing has no practical choice other than to assemble it identically to the way that the factory would, using the same materials. The family does not consider how much to spend on assembly because the example does not allow them any substitution possibilities. 16

182 frequency spectrum. Then the large incumbent has foreclosed its rival from access to lowfrequency spectrum. As the commenters hypothesize, moreover, suppose that the cost of providing equivalent service (summing both the price of spectrum and the cost of build out) using the high-frequency spectrum the firm has purchased is identical to the hypothetical cost the firm would have borne had it been able to purchase low-frequency spectrum at the auction price paid by the large incumbents and built out that spectrum to provide equivalent service. If the firm s only option once it purchased the high-frequency spectrum were to use it to provide equivalent service, then that is what it would do. But the firm may have another option for using the high-frequency spectrum: to spend less on build-out and offer service with less coverage, more limited building penetration, or lower capacity (which I will call targeted service). 27 That option may be its preferred method of using high-frequency spectrum, so when foreclosed from purchasing low-frequency spectrum, it would adopt this targeted build-out approach rather than spending more on build-out to provide equivalent service. 28 Although it would pay less for spectrum than it would have paid for low-frequency spectrum, its best 27 This outcome may arise even if the firm anticipates offering equivalent service eventually, but reaches it at a much later time because it may find it cost-effective to delay some of its investment decisions or spread them over a prolonged period of time. 28 There is nothing remarkable about the possibility that a mobile wireless provider would prefer to offer equivalent service if it can obtain a mix of low-frequency and high-frequency spectrum, but prefer to offer targeted service if it instead owns mainly high-frequency spectrum. The marginal benefit the firm receives from infrastructure investment likely differs depending on which spectrum it owns, and varies with how much investment the firm makes. In addition, it may take longer to build out to provide a given level of service using high-frequency spectrum than low-frequency spectrum, because it must employ many more cells to do so. See Fifteenth Competition Report 293 (citing National Institute of Standards and Technology (NIST) study to demonstrate that achieving similar geographic coverage requires nine cells at 2.4 GHz, four cells at 1.9 GHz, and one cell at 700 MHz). Moreover, the firm may anticipate that it would become more difficult to build or acquire cells as time goes by, for example, if delay means that best cell locations would be taken by other firms, zoning approvals for new sites would become harder to come by, or backhaul would become more costly. See, e.g. American Tower Corp., Annual Report (Form 10-K) (2012), available at Report/2012/12/31/t.aspx?t=XNYS:AMT&ft=10-K&d=a658a d603be47c235c8f78f2a ( Local zoning authorities and community residents often oppose construction in their communities, which can delay or prevent new tower construction, new antenna installation or site upgrade projects, thereby limiting our ability to respond to customer demand. In addition, zoning regulations can increase costs associated with new tower construction, tower modifications, and additions of new antennas to a site or site upgrades. ). The greater number of cells required to serve any given level of customer demand likely also would make the marginal cost of adding capacity higher for a provider relying largely on high-frequency spectrum. For all these reasons, a firm would not be expected to provide the identical level of service using high-frequency spectrum as it would choose if it had instead owned a mix of lowfrequency and high-frequency spectrum. 17

183 decision after acquiring the spectrum might not be to spend its spectrum savings on additional build-out. Under such circumstances, a large incumbent may be able to obtain or maintain market power by foreclosing rivals access to low-frequency spectrum. Absent foreclosure, the rivals would have purchased low-frequency spectrum and offered service comparable to what the large incumbents provide. When foreclosed, the rivals would instead purchase high-frequency spectrum and offer targeted service. If, as a result, the rivals provide less of a competitive constraint for the large incumbents, 29 those incumbents may be able to obtain or maintain market power, to the detriment of consumers. Here, a cap on low-frequency spectrum (in addition to an overall cap) would benefit consumers by preventing large incumbents from intentionally focusing their acquisitions on lowfrequency spectrum. Applying a separate cap to low-frequency spectrum holdings may offer a better way to achieve this competitive benefit than simply tightening a cap on overall spectrum holdings, moreover, because it targets the competitive problem and thus reduces the risk of preventing forms of spectrum aggregation that would confer scale economies without harming competition. C. Policy decision-making perspective It is likely worse, from an overall policy perspective, to fail to impose separate caps for different bands when such caps are justified than to impose such caps and later discover they are not necessary. If the Commission imposes separate caps but later concludes, notwithstanding the evidence presented by T-Mobile and others, that wireless services can be offered effectively using spectrum of any frequency, then the auction restriction would not make much practical difference to outcomes in mobile wireless services markets. In this scenario, by assumption, 29 Different firms may offer different quality levels in a competitive market, and this possibility does not by itself present a competitive problem. When many buyers view low-price, low-quality products as close substitutes for high-price, high-quality products, competition among the firms selling both types of products may prevent firms of either type from exercising market power. The concern here is that not enough buyers may view the products as close substitutes to protect the many buyers that prefer premium service from supracompetitive prices. Put differently, excessive agglomeration of low-frequency spectrum by large incumbents may induce a greater level of quality differentiation than would be provided by a competitive market, allowing the incumbents to exercise market power to the detriment of consumers. 18

184 spectrum frequency does not affect service quality, so firms that were blocked from increasing their holdings of low-frequency spectrum should be able to purchase other bands of spectrum to compensate. 30 By contrast, if the Commission does not impose separate caps for individual bands when such restrictions are warranted, it would allow wireless competition to be harmed. VI. Conclusion Spectrum auction rules to address excessive spectrum aggregation can foster competition in mobile wireless services. Auction rules with spectrum caps avoid costs, delays, and distortions in spectrum allocation that would result from relying on post-auction case-by-case review. Moreover, spectrum caps in auctions can encourage auction participation, so they have the potential to increase auction revenues. Separate caps for low-frequency spectrum can be beneficial even if wireless providers can overcome disadvantages of high-frequency spectrum with sufficient capital investment. Case-by-case reviews are more troublesome when applied to review auction outcomes than when applied to review secondary market transactions. When used in the latter case, moreover, they would be unlikely to undermine spectrum caps by encouraging speculative bidding in auctions. I certify that all statements made in this document are true to the best of my knowledge. Jonathan B. Baker March 12, Following the same logic, the fact that the two large incumbent providers most likely to be subject to national or regional caps on low-frequency spectrum oppose such a cap over-and-above their objection to spectrum caps is surprising given their claim that high-frequency bands could easily substitute for low-frequency bands. 19