December 26, 2008 CHICAGO Craig Strom Assistant Director of Engineering T-Mobile AWS Filter Implementation Progress Report 1.0 Abstract This report describes the testing and implementation of a prototype filter commissioned by T- Mobile as one possible solution to mitigating potential interference into BAS operations in the 2025-2110MHz BAS band. T-Mobile is one of several cellular operators who have recently bid on and won spectrum auctions for the 2110-2120MHz AWS block. These operations will cause interference into BAS receive systems at a level that will render the upper part of the BAS 2GHz spectrum unusable if not addressed. T-Mobile is the only company in this group that is taking aggressive action to ensure that their operations do not interfere with BAS operations in the adjacent band. Below you will find a discussion of the regulatory background that brings us here, a brief history of the testing performed before a solution was devised and the reasoning used to go forward with the solution tested here recently as well as results and recommendations on the actual lab and field tests conducted. All expenses (other than my time) involved with this effort have been paid for by T-Mobile. 2.0 Regulatory Background Some time ago, the FCC released CFR 47, Part 27 as part of the service rules for Advanced Wireless Service operations. This section of the Rules describes the service rules under which AWS companies must conduct business. Within those rules is Section 27.1133. This specific section details the requirement for AWS to protect existing BAS and CARS operations. It is included below for your reference. 27.1133 Protection of Part 74 and Part 78 operations. - AWS operators must protect previously licensed Broadcast Auxiliary Service (BAS) or Cable Television Radio Service (CARS) operations in the adjacent 2025-2110 MHz band. In satisfying this requirement AWS licensees must, before constructing and operating any base or fixed station, determine the location and licensee of all BAS or CARS stations authorized in their area of operation, and coordinate their planned stations with those licensees. In the event that mutually satisfactory coordination agreements cannot be reached, licensees may seek the assistance of the Commission, and the Commission may, at its discretion, impose requirements on one or both parties. While there were some issues with coordination early on in the process, T-Mobile takes this paragraph extremely seriously and has made consistent and comprehensive efforts to work with our industry and ensure that they do not interfere with our operations in the 2GHz band. 3.0 History of Testing and Investigation After the initial investigation of the impact of AWS operations on BAS receivers (thanks to Dane Ericksen for that work) it was determined that AWS operations would have a significant impact on ENG receivers. As AWS operations will not commence until the 2GHz relocation project has been completed in a particular market and Part 74 and 78 licensees will be using digital ENG almost exclusively, it was considered reasonable to confine the testing and measurement of interference to digital ENG operations.
T-Mobile attended a meeting on March 2, 2007 to begin a dialog with Chicago BAS licensees and the industry in general. While this meeting left us with a lot to learn, it set the stage for the T-Mobile staff to begin their education in field ENG and us to begin to understand the potential impact of this new service on BAS operations T-Mobile participated in several field tests over the next year with the Chicago broadcasters to begin to develop a practical understanding of how we use the 2GHz band. This also served to help educate us with regard to the workings of an AWS UMTS system. UMTS (Universal Mobile Telecommunications System) is the GSM implementation of the 3G wireless phone system. That is, UMTS is the modulation and transmission system used in the AWS service. During these early field tests (also in Chicago market using WLS-TV, WMAQ-TV and WBBM- TV receive sites), T-Mobile configured and operated several test AWS sites running simulations of fully loaded UMTS transmission systems. The extent of the degradation in BAS receiver performance was clear to the casual observer. These two field tests were followed by lab tests in April of 2008 conducted at the RF labs of Boeing in Seattle, WA. The lab tests used off the shelf digital ENG equipment that was configured and tested in commonly used SD and HD modes. The results of this test are available and have already been distributed to many industry representatives. The actual testing was attended by various T-Mobile and Boeing personnel as well as myself (representing ABC), Walter Sidas (representing CBS) and intermittently by Ron Diotte with KSTW the local CW station. The results of the measurements taken and performance parameters of the BAS radios used gave T-Mobile engineers the performance specifications necessary to commission several prototype band stop filters for the next phase of the project and, eventually, for field testing in Chicago. The lab testing also helped to quantify the two types of interference caused by an adjacent, high level interfering signal. It also provided the platform for T-Mobile to understand the structure of a typical post Nextel ENG receive site. The two types of interference are Out of Band Emission or, OOBE and Brute Force Overload or BFO. OOBE is the result of the AWS signal that is actually IN BAND for our BAS operations and results from the re-growth or skirts of the AWS signal from each transmitter and antenna in their system. BFO is the front end overload that results from an extremely high level signal in adjacent spectrum at close proximity to an ENG receiver. It was clear from the tests that the BFO was a huge problem. The OOBE interference might be an issue as well, but the BFO problem would have to be fixed first to begin to understand and test the OOBE issue. In the absence of another standard, we defined interference for T-Mobile with a simple test: If we can make a particular live shot work without presence of the AWS signal and we cannot when the AWS signal is present, then that is interference. We gave them 1dB of degradation as a limit not to be exceeded. Since we did not know which type of interference would have the limiting impact on the received signal, we tested BAS signal levels at threshold, mid-range and hot signal levels. At each level, AWS signal level was increased until the BAS receiver failed. We also tested 6MHz split channel operations with high level interfering signals from another digital BAS source. The interfering BAS signal was always on channel 7- and the desired signal was always on channel 7+. All of this testing was done on the new BAS band plan on channel 7. We always tested the most fragile modulation modes (HD) and usually included an SD robust mode for reference. All split channel testing was done with 6MHz COFDM or VSB signals.
The informal field testing and published lab testing led to the development and testing of 2 prototype band stop filters. The second test filter was quickly eliminated from the field test as it clearly demonstrated lower performance than the first. All references to the filter here refer to the first and higher performance filter manufactured by CMT. 4.0 Reasoning Behind the Receive Filter Mitigation Methodology There are several solutions specific to mitigating the two types of interference noted here. They are addressed separately here as is their impact on interference mitigation. The interference from OOBE can only be mitigated at the UMTS transmitter. This interference consists of the energy that the UMTS transmitter and antenna system generate that is out of band to the AWS system and in band for BAS (that is, energy that is developed below 2110MHz). AWS operators can only address this by increasing the out of band, or mask performance of their transmitter or adding a band pass filter (similar in function to the DTV mask filter) to the transmission path between the AWS transmitter and the transmit antenna. Interference from BFO can be addressed by several methods. First, transmit side antenna pattern control can be used to minimize the energy received from the UMTS antenna at the BAS receive site. Second, receiver side antenna pattern control can also be used to minimize the energy received from the AWS site at the BAS receiver input. And third, an RF filter rejecting the source of the interference can be employed at the receiver where the interference is experienced. The transmit antenna pattern control technique can be employed as needed on a case by case basis. In the end, it may be more realistic for AWS companies to employ this technique as a matter of course rather than re-engineering cell sites in close proximity to BAS receive sites. This technique consists of locating the BAS receive site at the nulls between AWS antenna panels and increasing shielding and beam tilt on the AWS antenna to minimize energy aimed directly at the BAS receive site. Unfortunately, it is not realistic to employ receive side antenna pattern control as it is the receive site goal to be able to receive from all bearings. Usually, a steerable antenna is employed and the AWS sites are ubiquitous in their location. The most effective method for reducing BFO interference in this case is clearly the BAS receive site band stop filter. From the lab tests, it was clear that this would be the most effective initial method of interference mitigation. In an ideal environment, this filter would be located after the BAS receive antenna and before any active components of the receive system. In our environment, this is not possible due the fact that we use low noise amplifiers in the feed assembly of our systems and the physical size (about 17 x3 x8 ) of the required filter. The impact of this limitation is that the filter performance is reduced by the gain of the LNA. While this is not ideal, there is no real choice to improve the situation. Another danger to this location for the filter is overload in the LNA (from high level AWS signals). The lab testing included LNA overload tests that showed the latest LNA hardware to have IP3 points much higher than necessary for avoiding LNA overload. With these facts in mind, a prototype filter was commissioned, designed and built by T-Mobile to best filter their signal from our band. The results of that effort are discussed below. 5.0 Filter Test Plan and Results
WLS-TV received the prototype filter in August 2008 for examination. Subsequent to that, the filter was tested with a test configuration detailed below in figure 1. To simulate the AWS environment a Nucomm CamPac transmitter was configured to behave much like a UMTS transmitter. The spectral and energy characteristics are not identical to the UMTS transmitter, but gave a reasonable approximation for initial filter tests (and, it was available ). This transmitter was configured for a 5MHz bandwidth and tuned to 2112.5MHz, the center of the A1 AWS channel. The spectrum plots (shown below as figures 2 through 6), detail the performance of the filter on the bench. From these plots if is clear that the prototype band stop filter has very high performance characteristics and is a good candidate for field testing in an actual UMTS /BAS environment. Note that the insertion loss of this filter is 1dB. Figure 1, test configuration
Figure 2, prototype filter response
Figure 3, AWS simulator pre-filter
Figure 4, AWS simulator and BAS (6MHz, COFDM) signals pre-filter
Figure 5, AWS simulator and BAS signals through test filter
Figure 6, AWS simulator through test filter The results in figures 2 through 6 were measured on channel at 2110MHz. These results were sufficiently positive to merit moving to the field test phase with this filter. For the field test, T-Mobile activated 12 cell sites with UMTS modulation and traffic simulations at maximum loading covering about 180 degrees around the WLS Lake Zurich ENG receive site. These cell sites were located in southern Lake County, Illinois. T-Mobile operated these sites under AWS license AWS-CMA003A. The predicted worst case bearing (to the closest cell tower) was 92 degrees. This bearing was used for the first day of testing and produced a - 70dBm (with no LNA) level at the ENG receiver (tuned to channel 7+). Unfortunately, the field crew working on the sites had neglected to rotate the panel on this adjacent site to point to the ENG receive site (we suspected a problem as -70dBm was markedly lower than predicted). The second day, we looked for higher level UMTS signals and found the 2 nd closest site that produced a -60dBm at the ENG receiver (with no LNA) at 169 degrees from the ENG receive site. All of the previous day s tests were repeated at this new bearing. It would be reasonable to predict that this level could be higher in practice as it would be reasonable to assume that cell towers might be located closer to the ENG receive site of interest. In that event, additional mitigation might be required to achieve results that are acceptable. For the record, this ENG receive site uses a dual band (2 & 7GHz), quad polarity Radiowaves Proscan III receive antenna with LNA bypass mounted at approximately 320 above ground.
All of the following plots are captured at the 70MHz 2 nd IF output of a Nucomm CR6D ENG receiver. The signal levels at 2GHz were too low to capture meaningful spectrum plots, so the 2 nd IF output of the Nucomm CR6D was used. The most telling statistic is the simple observation of whether or not the receiver was locked and stable. Each plot notes the ENG modulation parameters and salient environmental statistics. We tested various worst case, but typical ENG operating modes. Modes that were considered less likely to fail than others were not tested. For example, 6MHz COFDM SD was tested on a plus offset, but not on a minus offset. In the same way, 8MHz COFDM was tested on channel 7 center at HD data rates, but not at SD. It was planned that if any of the more fragile modes of operation had failed the test, the less fragile modes would be tested, however, none of those modes failed so the tests were not stepped back to these less fragile modes. The first group of plots (Fig 20 Fig 27) is the dual adjacent channel tests, or sandwich tests. These tests used a very strong interfering BAS signal (6MHz pedestal, channel 7-, COFDM using QPSK modulation) as an undesired signal and various desired signals using 6MHz pedestals on channel 7+. The AWS signal provided the upper source of interference. In each suite, the first three tests were conducted with the prototype filter installed, the fourth with it removed. The first three tests were repeated with the LNA bypassed, at low gain and at high gain. The second group of plots (Fig 28 Fig 39) is the single adjacent tests. These tests used only the AWS signal as an interfering signal. These tests included both 6MHz pedestals on channel 7+ as well as 8MHz pedestals on channel 7 center. Again, each test configuration was run with the filter in and the LNA off, at low power and at high power. The filter was then removed and the LNA bypassed for the fourth test. We did not conduct tests with the filter out and the LNA on as the receiver had already failed with it off and shots with additional amplification would have also failed yielding no additional information. The final test (Fig 40) was conducted with a high level BAS signal to ensure that nothing unexpected would occur at higher BAS signal levels. Only the most fragile mode was tested as our test time was quickly coming to an end and further improvement was not anticipated with different configurations. Each plot is explained in the notes for the plot and summarized in the recommendations that follow. In each test, the threshold of the desired signal is established without the UMTS signal present. In the case of the double adjacent channel tests, the threshold is established with the lower adjacent undesired BAS signal present. Then the undesired UMTS signal is applied and the measurements are taken. An additional 1dB of signal was added to the desired signal after establishing the threshold to ensure a locked and stable signal.
Fig. 20 Double Adjacent Channel Test Tx Power mode: High Modulation: 8T VSB (HD) LNA: Bypass Rx Receive Level: -74dBm AWS Filter: IN Receiver LOCKED & STABLE
Fig. 21 Double Adjacent Channel Test Tx Power mode: High Modulation: 8T VSB (HD) LNA: Low Rx Receive Level: -63dBm AWS Filter: IN Receiver LOCKED & STABLE
Fig. 22 Double Adjacent Channel Test Tx Power mode: High Modulation: 8T VSB (HD) LNA: High Rx Receive Level: -49dBm AWS Filter: IN Receiver LOCKED & STABLE
Fig. 23 Double Adjacent Channel Test Tx Power mode: High Modulation: 8T VSB (HD) LNA: Bypass Rx Receive Level: -74dBm AWS Filter: OUT Receiver: NOT LOCKED
Fig. 24 Double Adjacent Channel Test Modulation: QPSK, 2/3 FEC, 1/32 Guard Interval (SD) LNA: Bypass Rx Receive Level: -85dBm AWS Filter: IN Receiver: LOCKED & STABLE
Fig. 25 Double Adjacent Channel Test Modulation: QPSK, 2/3 FEC, 1/32 Guard Interval (SD) LNA: LOW Rx Receive Level: -74dBm AWS Filter: IN Receiver: LOCKED & STABLE
Fig. 26 Double Adjacent Channel Test Modulation: QPSK, 2/3 FEC, 1/32 Guard Interval (SD) LNA: HIGH Rx Receive Level: -60dBm AWS Filter: IN Receiver: LOCKED & STABLE
Fig. 27 Double Adjacent Channel Test Modulation: QPSK, 2/3 FEC, 1/32 Guard Interval (SD) LNA: Bypass Rx Receive Level: -84dBm AWS Filter: OUT Receiver: NOT LOCKED
Fig. 28 Single Adjacent Channel Test Modulation: 8T VSB (HD) LNA: Bypass Rx Receive Level: -84dBm AWS Filter: IN Receiver: LOCKED & STABLE
Fig. 29 Single Adjacent Channel Test Modulation: 8T VSB (HD) LNA: Low Rx Receive Level: -72dBm AWS Filter: IN Receiver: LOCKED & STABLE
Fig. 30 Single Adjacent Channel Test Modulation: 8T VSB (HD) LNA: High Rx Receive Level: -59dBm AWS Filter: IN Receiver: LOCKED & STABLE
Fig. 31 Single Adjacent Channel Test Modulation: 8T VSB (HD) LNA: Bypass Rx Receive Level: -84dBm AWS Filter: OUT Receiver: NOT LOCKED
Fig. 32 Single Adjacent Channel Test Channel: 7 Center Modulation: 16 QAM, 7/8 FEC, 1/32 Guard Interval (HD) Pedestal: 8MHz IF Filter B/W: 8MHz LNA: Bypass Rx Receive Level: -86dBm AWS Filter: IN Receiver: LOCKED & STABLE
Fig. 33 Single Adjacent Channel Test Channel: 7 Center Modulation: 16 QAM, 7/8 FEC, 1/32 Guard Interval (HD) Pedestal: 8MHz IF Filter B/W: 8MHz LNA: Low Rx Receive Level: -74dBm AWS Filter: IN Receiver: LOCKED & STABLE
Fig. 34 Single Adjacent Channel Test Channel: 7 Center Modulation: 16 QAM, 7/8 FEC, 1/32 Guard Interval (HD) Pedestal: 8MHz IF Filter B/W: 8MHz LNA: High Rx Receive Level: -63dBm AWS Filter: IN Receiver: LOCKED & STABLE
Fig. 35 Single Adjacent Channel Test Channel: 7 Center Modulation: 16 QAM, 7/8 FEC, 1/32 Guard Interval (HD) Pedestal: 8MHz IF Filter B/W: 8MHz LNA: Bypass Rx Receive Level: -86dBm AWS Filter: OUT Receiver: LOCKED but not STABLE
Fig. 36 Single Adjacent Channel Test Modulation: QPSK, 2/3 FEC, 1/32 Guard Interval (SD) LNA: Bypass Rx Receive Level: -92dBm AWS Filter: IN Receiver: LOCKED & STABLE
Fig. 37 Single Adjacent Channel Test Modulation: QPSK, 2/3 FEC, 1/32 Guard Interval (SD) LNA: Low Rx Receive Level: -80dBm AWS Filter: IN Receiver: LOCKED & STABLE
Fig. 38 Single Adjacent Channel Test Modulation: QPSK, 2/3 FEC, 1/32 Guard Interval (SD) LNA: High Rx Receive Level: -68dBm AWS Filter: IN Receiver: LOCKED & STABLE
Fig. 39 Single Adjacent Channel Test Modulation: QPSK, 2/3 FEC, 1/32 Guard Interval (SD) LNA: Bypass Rx Receive Level: -87dBm (AWS signal w/o filtering increases this from -92) AWS Filter: OUT Receiver: NOT LOCKED
Fig. 40 High Level BAS Test Modulation: 8TVSB (HD) LNA: Bypass Rx Receive Level: -54dBm AWS Filter: IN Receiver: LOCKED & STABLE The 1 st IF noise floor increased by 8 to 10 db with this hot BAS signal.
6.0 Recommendation based on Results The most telling statistic from these tests was that, in every case (except the high level test where the receiver was locked at all times), the BAS receiver was either unlocked completely or taking severe errors with the prototype filter out of the circuit, and, conversely, it was locked in every case with the filter in the circuit. It should be noted that BAS receive level had to be increased by approximately 1 db in several cases to make up for the 1dB insertion loss of the filter. As the receiver signal strength meter could not resolve down to tenths of a db, this variance could be up to + or 1 db. Our impression at the time of the adjustment was that it was just under 1dB, but that depends on the exact level of the BAS signal relative to the BAS threshold. It should be noted that no testing of the upper DRL channels was done. It is very likely that the presence of high level UMTS signals and BAS signals will render the upper DRL channels unusable. The eventual presence of wireless services immediately below the lower DRL channels will likely render those unusable as well. This is a subject for additional testing as DRL equipment becomes available. In summery, we tested and successfully mitigated AWS interference from a BAS receive system at AWS UMTS levels up to -60dBm (at the BAS receiver) using a prototype band stop filter commissioned by T-Mobile. Higher levels were not tested and may or may not require additional mitigation techniques. Those additional techniques might include specifying minimum distances around ENG receive sites and/or AWS antenna orientation and/or shielding optimized to minimize the UMTS signal at the BAS receive antenna. These additional efforts would be intended to keep the UMTS signal level beneath -60dBm (or a higher level determined in the future). Additional testing should be undertaken to determine the maximum level of UMTS signals that the filter will mitigate. Respectfully Submitted, Craig Strom WLS-TV Chicago