Abstract INTRODUCTION Background to the Problem Purpose of the Study...9

Transcription

1 Mpeg-4 over CATV Table of Contents Abstract INTRODUCTION Background to the Problem Purpose of the Study Statement of the Problem Methodology Outline of Remaining Thesis Chapters MPEG COMPRESSION AND SIGNAL DELIVERY Introduction MPEG-1 and -2 Compression Systems CATV Networks Summary of Findings MPEG MPEG-4 Compared MPEG-4's New Paradigm MPEG-4 Tools The MPEG-4 Delivery Layer MPEG-4 Logic Broadcast Scenario The IM1 Implementation Summary MODELS AND EXPERIMENTS Verification Model (VM) Core Experiments Summary...38

2 Mpeg-4 over CATV METHODOLOGY, RESULTS, COMMENTS Methodology Results and Conclusions Comments...42 Bibliography...44 Appendix A-1: Tables...47 Appendix A-2: Figures...48 Appendix B-1: Acronyms and Abbreviations...49 Appendix B-2: Glossary...50

3 Mpeg-4 over CATV Abstract This dissertation study describes and presents an alternative method for delivering digital video, the MPEG-4 compression format, through the use of an analog cable network for direct viewing on television sets. The purpose of the study is to propose this technology as an alternative to the present cable network technologies in Uruguay and elsewhere in South America. Delivery standards that might be used over such a CATV network include DVB-C (digital video broadcast for cable), and IP (internet protocol) networks. The study compares the current technologies to determine which would be the best way to provide such delivery over CATV networks. It is the argument of this study that the delivery of MPEG-4 format streaming video is no different than the delivery of any type of digital signal: in order for the signal to be delivered, it must pass through an analog medium-copper wire, coaxial cable, fiber optic cable. The dissertation makes the case that MPEG-4 is a feasible alternative technology which should be considered for new CATV applications in this region. This is from a technical point of view, as Leonardo Chiariglione 1 points out about this study, when it comes to implementati0n, there are other elements that should be taken into account. Keywords: MPEG-4, CATV, Broadcasting, Streaming Media. 1 I will ask for permission to include the person s name.

4 Mpeg-4 over CATV INTRODUCTION 1.1 Background to the Problem The compression of data so that it can be transmitted over wide area networks (WANs) economically and rapidly is a well-established digital technology, and an ongoing concern in a world where information is a prime commodity. The need to compress arises from the fact that the more complex the information, the greater the amount of electronic data (in terms of digital zeros and ones) needed to represent it- whether for data storage and retrieval, or for transmission purposes. For example, plain text requires 8bits/character (in the order of 20kbits/page); music such as we find on a CD needs about 1500 kbits for every second (or 1500Kps); but full-motion uncompressed TV video signals require more than 200 Mbits for every second of transmission 2. While the need to compress data is obvious in the storage and retrieval of information, it is clearly in the timely transmission of such data that the majority of the problems are encountered. In fact, without compression, the same digital video signal would take up five times as much bandwidth as a comparable analog signal. In terms of a TV signal, this means that it would take five TV channels to carry one channel of digital television (Parsons & Frieden, 1998). In an ideal world, we could send an infinite amount of information given enough time and bandwidth. In the real world, however, both time and bandwidth are restricted. Thus, straight-connection telephone modems for example are restricted to transmitting a maximum of 56Kps-and that is under ideal conditions which can never occur in actual telephone lines. ADSL (asymmetric digital subscriber line) modems can reach downstream speeds of 8Mps and upstream speeds of 1Mps while cable modems can have the potential of up to 30Mps of downstream bandwidth (but this is typically shared among several thousand users). Thus, even the fastest cable modem being used by only one person would take seven seconds to download one second of television-quality video while the typical cable modem would download that one 2 Estimated using the CCIR-601 standard.

5 Mpeg-4 over CATV second of video in the region of several hours. Table 1.1 below shows some of the access and delivery systems and networks and their potential rates. Service/Network POTS (plain old telephone system) ISDN (Integrated Services Digital Network) ADSL (Asymmetric Digital Subscriber Line) Rate Kps Kps Mps (downstream) Kps (upstream) Mps 20-40Mps Nx51.84Mps VDSL (Very High Rate DSL) CATV (Cable Television) OC-N/STS-N (Optical Cable-Number of times the single link bandwidth/synchronous transport protocol-number of times the single ink l bandwidth Ethernet 10Mps Fast Ethernet 100Mps Gigabit Ethernet 1000Mps FDDI (Fiber Distributed Data Interface) 100Mps (wireless) 1,2,5.5, and 11Mps in 2.4 GHz band a (wireless) 6-54Mps in 5 GHz band Table 1.1: Networks and Network Services 3 According to Chen (2002): "Without question, digital video compression is the most important enabling technology for modern video communication" (p1). Some of the advantages of compressed digital video over analog include: Lower video distribution costs. Better video quality and signal security. Potential for interactivity. When it comes to audio and video compression, the standards have been set by the Moving Pictures Experts Group (MPEG) under the auspices of the International Standardization Organization (ISO). The first set of standards issued were MPEG-1 (ISO/IEC 11172), which was concerned mainly with compression issues for storing and retrieving video and audio rather than communication and transport. The standard has been used mostly for CD-ROM storage and for audio, especially in the form of MP3. As well, it has been used for low resolution video 3 Gibson, 2001a, p3.

6 Mpeg-4 over CATV compact discs, providing about one hour of 320x240 video on a 650MB CD-R or CD-RW disc. MPEG-2 (ISO/IEC-13818), which includes the MPEG-1 standards (traditionally the MPEG standards are backward compatible so as not to make previous hardware and software obsolete), has a systems component which "defines program streams suitable for applications over a reliable medium, and transport streams suitable for networked delivery" (Kalva, 2001, p3). The standard has become very popular and used for digital satellite transmission as well as DVD video discs (two hours of video on a 4.7GB disc). In fact, MPEG-2 has worked so well that work on an MPEG-3 set of standards for High Definition TV needs was dropped and rolled back into MPEG-2: MPEG is used in a number of substantial commercial applications, including playback of media from disk storage, digital broadcast of audio-visual programming over a variety of channels, point-to-point switched connections for delivery of digital audio-visual material, high definition television, and networked multimedia (Lookabaugh, 2001, p116). MPEG-4 (ISO/IEC-14498) differs significantly from MPEG-1 and MPEG-2. Rather than compressing by using an entire video frame or field, MPEG-4 uses a layered approach, separating the scene into foreground and background. For example, if you had a person jogging in the foreground and a relatively static city scene in the background, MPEG-4 would treat them as two different layers and use different compressions for each. One of the problems with MPEG-2 compression of rapidly moving objects when viewed over a digital network, for example, is that they sometimes tend to break up or produce "artifacts," defined as "features or elements of the input that are not truly part of the information" (Symes, 1998, p5). Because MPEG-4 would compress the background more strongly than the foreground, you would get a high-quality image without artifact noise. The question many are asking is: exactly how high quality will the broadcast image be? Gomez (2002) says: "According to MPEG-4 proponents, at a rate of about 300Kps, you'll be theoretically able to see VHS-quality images playing back in real-time. The 300Kps data rate is possible with today's DSL or cable modem speeds" (p66).

7 Mpeg-4 over CATV There is one element that has not been discussed thus far and which is crucial no matter what type of compression is used-and that is the machinery to encode a signal at one end and decode it at the other. This equipment must be able to "pack" a signal before it is sent and "unpack" it before it is viewed. Figures 1.1 and 1.2 show simplified schemata for MPEG encoders and decoders respectively. Figure 1.1: Simplified MPEG encoder Figure 1.2: Simplified MPEG decoder

8 Mpeg-4 over CATV In order for a compression system to work, the encoder and decoder must be compatible. The original MPEG designers made sure of this compatibility by coming up with a definition of the signal between the two devices rather than a definition of the devices themselves. Watkinson (2002) says: Developing compression tools is rather like an arms race. Engineers build a simple decoder that works well, but occasionally it encounters a particular type of picture that it can't compress. So the engineers develop a new tool to fix that. The result is that the encoder encounters uncompressible pictures less often, but it still happens. So the engineers keep developing new tools to keep up with compression difficulties (p2). The latest result of this process is MPEG-4 (with MPEG-7 already being developed-called a Multimedia Content Description Interface "concerned with the interpretation of information in such a way that it can be used or searched by devices or computers" (Gibson, 2001b, p.127)). MPEG-4 is a long way from MPEG-1 when it comes to complexity and the types of objects it can handle. At its most sophisticated, MPEG-4 can compress and transmit three-dimensional virtual objects as meshes which can then be recreated at the other end using a "map" of the object's surface or texture. As well, the object can be recreated from a chosen viewpoint-and that viewpoint could be that of the viewer in an interactive system. Finally, because of the way MPEG-4 is designed, it is separate from the delivery system and can be used by an expanding variety of these systems, including UDP (User Datagram Protocol) over IP, PSTN (Public Switched Telephone Network), ATM (Asynchronous Transfer Mode), and MPEG-2 streams. However, there is a price to pay for such complexity-and that is a rise in the complexity of the encoder and decoder: MPEG-4 probably represents the practical limit in coding complexity. Although it can reduce a moving image to a few vectors, it does require the decoder to be a powerful graphics-rendering engine. And although the object-based tools of MPEG-4 are very efficient, they are easily applicable only to computer-generated images. In principle, an encoder could be built that would take in real video from a natural image and dissect it into objects, but this would be a very complex process (Watkinson, 2002, p5-6).

9 Mpeg-4 over CATV Purpose of the Study The rationale behind this paper and the purpose of the study is an attempt to propose some alternative methods for delivering digital video in MPEG-4 compression format through the use of an analog cable network. The form of video is not to be delivered for storage purposes and later retrieval, but rather for direct viewing on a television set. This means that a real-time factor is involved, in the sense that the time taken for delivery cannot exceed the speed at which the video is streamed. Available standards that might be used over a CATV network for such delivery include DVB-C (digital video broadcast for cable), and IP (internet protocol) networks. The study will examine these methods to determine which may prove to be the best way to provide such delivery over CATV networks. Basically, the delivery of MPEG-4 format streaming video is no different than the delivery of any type of digital signal: in order for the signal to be delivered, it must pass through an analog medium-copper wire, coaxial cable, fiber optic cable. Even wireless signals must be sent over an analog medium. Table 1.2 below indicates how digital and analog are used within any cable television systems. Cable television system components Satellite feed to head end Return path pay-per-view message from set-top box Fiber feed for importing remote TV channels NTSC (National Television System Committee), PAL (phase alternate line) or SECAM (sequential width memory) television Cable modem (also on phone line) Hybrid fiber coax (HFC) signal distribution Signal characteristics Digital or analog microwave signals over the air on a microwave carrier Digital information modulated onto a return carrier dedicated to the purpose Analog and digital signals light wave modulated onto an optical fiber Analog video modulated onto a carrier and transmitted over the air on a cable or fiber Digital base band modulated onto a burst carrier Analog and digital signals transmitted as analog signals. Table 1.2: Digital and Analog Use Throughout Cable Television Systems 4 4 Thomas & Edgington, 1999, p3.

10 Mpeg-4 over CATV Put another way, the purpose of this study is to determine if the MPEG-4 compression format is compatible and consistent with streaming video delivery on conventional cable television network. 1.3 Statement of the Problem When it comes to MPEG-4 streaming video, what needs to be done is simple the implementation of a compatible cable head-end encoding/streaming system at one end and a customer side decoding solution at the other. One of the things to be noted about the MPEG compression algorithm is its asymmetrical nature-the compressing side of the equation is much more complex (in terms of both hardware and software) than the decompressing side-and this is as it should be when it comes to transmitting signals to a customer: For cable television this means the head end sources of programming material, such as digital center from a satellite or fiber link, do all the compression work. Neither the head end nor the receiver of the compressed material has to be packed with processing hardware and software. The subscriber's set-top box can be lower cost, standardized, and simple (Thomas & Edgington, 1999, p ). The asymmetry is also true of MPEG-4. However, as noted above, the MPEG-4 toolbox is itself becoming very complex (perhaps reaching the limits of this type of compression in relation to cost). While Kalva (2001) is describing the use of the MPEG-4 format for multimedia audio-visual applications, what is said here can just as easily apply to attempts to transport this type of content over cable to the end user: With object-based audio-visual representation, the presentations can contain many different media types and it is impractical to have a terminal with hardware decoders Terminals supporting object-based presentations would have to include software decoders and even programmable processors for efficient decoding With object-based multimedia content and the support for user interaction, the user terminals are becoming more and more complex (p9).

11 Mpeg-4 over CATV Thesis Argument: The thesis argument being made in this paper can be stated thus: "It is possible to deliver streaming video in MPEG-4 format through an analog CATV network so that an end-user can watch the video in real time without having to download it onto some type of storage device. Not only must the end-user be able to view the video but he/she should be able to make use of MPEG-4's object-oriented compression toolbox to achieve full interactivity (otherwise, MPEG-2 would continue to be sufficient). In order to do that, a solution needs to be found to resolve the issues connected with the complexities of both the encoding/stream at the cable head-end and decoding at the user end. This paper is an examination of potential solutions to this problem." 1.4 Methodology The testing of the above thesis statement is to be done through a combination of qualitative and quantitative methods. These methods include: An examination and analysis of the pertinent practical and theoretical literature written on the subject of MPEG compression, cable network configurations, analog and digital signal manipulation, and encoding-decoding technology. This includes technical material in electronics, computer and audiovisual engineering such as video-on-demand broadcasting protocols (Carter, Long, & Pâris, 2001), and future telecommunication networks (Kazovsky, Khoe, & von Deventer, 2001). An examination of previous studies and experiments performed in this area-with specific emphasis on analogous MPEG-2 uses within the cable TV network areas. For example, U.S. cable operators who converted to two-way hybrid fiber coax to implement the MPEG-2 protocols already have a leg up. As Yoshida (2000) indicates: "The cable industry's initial

12 Mpeg-4 over CATV interest in the MPEG-4 standard is as a low-bit-rate streaming media codec. Cable operators could broadcast low-bit rate MPEG-4 streams from a head-end using MPEG-2 transport, or MPEG-4 could offer point-to-point delivery of streaming AV via Internet Protocols" (p2). As well, the study will examine specific experiments using MPEG-4 formatted content with a variety of encoding and decoding equipment as carried out by the MPEG group, designated experts and labs outside the group, and accepted proposals from outside laboratories. It has been suggested that one of the problems with MPEG-4 is that it was created with convergence markets 5 in mind and designed to work within those ever-expanding markets. This "abstractness" and expandability could prove a problem for singular practical purposes. When it comes to cable transport, it is important that some standard be reached in terms of object complexity, for example: "For cable, a standard streaming video profile would ensure that the settop could detect incoming MPEG-4 objects, to assure it could download software and to establish that its CPU could decode a number of objects. The box could then render those objects in software and overlay them on MPEG-2" (Yoshida, 2000, p3). 1.5 Outline of Remaining Thesis Chapters The second section of this paper consists of an examination and analysis of literature on MPEG-1 and MPEG-2 standards and cable network transport abilities with specific reference to the sending of compressed digital signals over analog networks and the encoding/decoding hardware/software required. The next section (Chapter Three) consists an evaluation of the MPEG- 4 history and standards with specific attention paid to the Delivery Layer within 5 Telecommunications, computing, and TV/film/entertainment worlds.

13 Mpeg-4 over CATV MPEG-4 and how it affects the potential transport of MPEG-4 digital video as well as specific present studies and research conducted on MPEG signals transmitted over cable networks. This is followed in Chapter Four by specific present-day studies and research conducted on MPEG signals over cable networks, as well as the results of experiments with streaming video in MPEG-4 format. The final section (Chapter Five) consists of a summary of the results, a set of conclusions, an appraisal of the offered solution to the problem of transmitting streaming MPEG-4 video over analog cable networks, evaluation of the feasibility of such a network technology in Montevideo and comparable cable television networks, and presentation of suggestions for future research in this area.

14 Mpeg-4 over CATV MPEG COMPRESSION AND SIGNAL DELIVERY 2.1 Introduction The material in this section consists of a literature review of specific subject matter-the MPEG standards for video coding and compression, and the delivery systems-in this case, analog CATV networks. The purpose is to examine the compatibility of the two for video transmission as currently configured-with specific reference to the MPEG-2 standard presently in use. The materials were obtained from a variety of sources including books, manuscripts, journals, articles, Internet web sites, and the InfoTrac Expanded Academic ASAP database. The chapter consists of four sections, including the introduction. The other three sections are: MPEG-1 and -2 Compression Systems, Analog CATV Networks, and Summary of Findings. 2.2 MPEG-1 and -2 Compression Systems As Solari (1997) points out, video compression systems such as JPEG and the MPEG family do not work on one single compression technique but rather on a group or family of techniques. This allows developers to choose among the techniques to find the one that best suits their particular applications. MPEG-1 was created with the intent of finding ways to encode audio and video onto a compact disc with a constant bit rate of 1.5 Mbps. Specifications for MPEG-1 included (Lakhani, 1996): 360-pixel horizontal resolution; 240-pixel vertical for NTSC (National Television System Committee-North America); 288 for PAL (Phase Alternating Line) and SECAM (Sequential Couleur Avec Memoire); 30-Hz frame rate for NTSC; 25 for PAL and SECAM; 24 for film. The aim of MPEG-2 was to define a standard applicable for true TV resolution-four times what MPEG-1 standards could provide. MPEG-2 uses a six-

15 Mpeg-4 over CATV layer hierarchical structure in order to break down the data into units that can be easily decoded. Table 2.1 below shows what these layers consist of. Syntax Layer Sequence layer Group-of-pictures layer Picture layer Slice layer Macro block layer Block layer Functionality Context unit Random-access unit: video coding Primary coding unit Resynchronization unit Motion compensation unit DCT unit Table 2.1: Layers of MPEG-2 Video Bit-Stream Syntax 6 The first step in producing an audio/video signal that takes up the least amount of bandwidth while maintaining an acceptable standard of quality lies in the encoding of the original analog signal. The reasons for the use of MPEG-2 encoding in DVB (digital video broadcasting) are: Its support of different video qualities up to HDTV (High Definition TV) programming; Its high flexibility. According to de Bruin & Smits (1999): The encoding system reduces the bandwidth by subtracting successive parts of the digital video signal. In case these successive parts are equal (e.g., the television screen is only showing one color), no information is encoded. At the decoder the subtracted information is added to the next part of the signal again to reconstruct the complete digital video signal. The MPEG-2 standard prescribes a subtraction taking pace at the same frequency as that at which a picture on the television screen is updated (p147). The MPEG-2 decoding system basically reverses the process used in encoding. One of the things that MPEG-2 does is allow for spatial scalability. This means that the resolution of a picture on a TV screen is not fixed but can be optimized-through the use of a digital video signal that includes both a basic resolution and a high resolution. 6 Solari, 1997, p221.

16 Mpeg-4 over CATV The other thing that makes MPEG-2 different is the use of "levels" and "profiles" within its set of standards. Thus, rather than simply one standard, MPEG-2 allows for four data formats or levels and five different profiles, which each having its own set of compression tools. The four levels range from low-definition television (VCR quality), standard-definition (PAL, SECAM, and NTSC), enhanceddefinition, and HDTV. The five profiles consist of ever-increasing complexity: simple, main, SNR (signal-to-noise ratio) scalable, spatial scalable, and high. Of the 20 possible combinations of levels and profiles, 11 were selected for use in the MPEG-2 standards. Table 2.2 below lays out the 11 combinations or "MPEG-2 conformance points" (de Bruin & Smits, 1999, p150). Profiles Low-Level Main-Level High-1440 Level High-Level Simple 720x576 (15Mbps) Main 352x288 (4Mbps) 720x576 (15Mbps) 1440x1152 (60Mbps) 1920x1152 (80Mbps) SNR scalable 352x x576 (4 or 3Mbps) (15 or 10Mbps) Spatial 1440x1152 or 720x576 scalable High 720x576 or 352x288 (20 or 15.40Mbps) (60 or 40.15Mbps) 1440x1152 or 720x576 (80 or 60.20Mbps) Table 2.2: Layers of MPEG-2 Video Bit-Stream Syntax x1152 or 960x576 (100 or 80.25Mbps) In order for MPEG encoded video signals to be successfully transmitted, they must contain three elements: audio information, video information, and information to provide support for the first two elements. MPEG-2 provides what is called the "MPEG-2 systems" to provide for the comprehensive transmission of these signals. This is in the form of packetizers which produce a stream of packets (packetized elementary stream of PES) for each of the three elements. These PESs then go through a multiplexer (MUX) to produce a standardized data stream known as the transport stream (TS). Figure 2.1 below shows a functional representation of the MPEG-2 systems. 7 de Bruin & Smits, 1999, p151.

17 Mpeg-4 over CATV Figure 2.1: Functional Representation of the MPEG-2 Systems 8 Also important is the fact that MPEG-2 is backward compatible with MPEG-1 through its program stream. This, along with its levels and profiles, makes MPEG-2 generic. De Bruin & Smits (1999) conclude: The MPEG-2 standard provides a "tool kit" of compression and transmission techniques. For any application, users must choose which tool to use This should give service providers and/or manufacturers an incentive to distinguish themselves. The user could benefit from this, because a variety of services with distinct quality could be placed at his our her disposal (p.163). Because of its multiple tool kit and its ability to code interlaced video, MPEG-2 has made itself useful in many applications, including digital terrestrial broadcasting, digital satellite TV, digital cable TV, DVD and others. In fact, several TV networks (including BBC and ITV in Britain) and satellite companies now broadcast some of their programming in MPEG-2 coded form (Ghanbari, 1999). Wiseman (n.d.) points out that universal standards for the definition of video compression are obviously needed; however, the crunch comes when these standards are to be implemented-at which time cost factors come into play. For example, it would not be worthwhile for an end user to have to acquire several 8 de Bruin & Smits, 1999, p153.

18 Mpeg-4 over CATV set-top boxes in order to decode and play video signals coming from several different and incompatible compression and transmission codecs. Part of the decision to select the MPEG group as the standard for digital video compression and transmission is because MPEG is cross-platform. As well, "The MPEG standard allows application developers to create customized trade-offs between playback rate, quality, storage capacity and cost" ("Professional Digital Video Networking," n.d., p4). 2.3 CATV Networks CATV networks consist of shared cable systems which use a so-called tree-and-branch topology so that one centralized signal can be sent to a multiple number of households. Originally, all these systems were coaxial cable; more recently, many of the systems of dual systems made up of fiber cable and coaxial cable or what is known as a hybrid fiber coax (HFC). The reason for using fiber when replacing coaxial is that fiber works much better when signals need to be sent longer distances. Also, fiber cable allows for increased bandwidth so that expansion does not require the updating of physical plant. In essence, the CATV system broadcasts the same type of radio frequency (RF) signals one would get through a TV antenna, except that they are carried in a cable (Sheldon, 2001). However, cable differs from through-the-air signals in significant ways, one of these being the fact that it can support (at least in theory) an almost infinite bandwidth expansion-thus the "broadband" name. However, in practice, because the line itself serves as a resistance to the signal, the signal can only travel so far before it loses its strength. Thus, the need for signal amplifiers at various distances. Originally in place because many TV sets were not "cable ready," the set-top box or converter is now used mainly to provide more advanced services such as pay per view or programming beyond the basic cable offerings (Parsons & Frieden, 1998). As well, until recently, very few television sets could accept digital video signals directly. These TV sets need a settop decoder box similar to the original set-top converters.

19 Mpeg-4 over CATV applications being foreseen: Not all cable systems are capable of delivering all the digital For cable systems that do not have sufficient capacity and lack the capital to upgrade, a hybrid system that combines the cable with satellite delivery may be used. The existing coaxial system will continue to carry analog programming while digital signals are delivered via one of the DBS [direct broadcast satellite] services and the two signals mixed in the customer's set-top box (Parsons & Frieden, 1998, p87). As a general rule, serial digital signals cannot be transmitted directly over a coax network. However, with the right transceivers and/or modems, an analog line can be converted into digital. For example, a digital feed (most likely in MPEG-2 compressed form) is received at the head end. There, the signal goes through a modulator which converts it into a modulated waveform (analog signal) to an up-converter. The up-converter translates the modulated signal to a selected CATV channel and this digital signal is then sent through the coax network. According to Maxwell (1999): As this digital signal cannot be applied directly to an analog television, a set-top box must be supplied at the subscriber's premises. It converts digital video back to analog by down-converting the RF carrier to base band, demodulating it to recover the original serial data stream, then decoding the data. Decoding here involves separating the audio and video components, decompressing the audio and video signals, and combining them again into a single NTSC-compatible analog signal for direction connection to the television (p148). One of the keys to the successful transmission of digital video signals through an analog medium is the reliability of the cable head-end. Presently, this head-end must be able to deal with both digital and analog signals-at least until there is a phase out of analog signals altogether. This makes head-ends "much larger and more complicated, requiring more maintenance, testing, and record keeping" (Bartlett, 2000, p188). Despite compression techniques, one of the major problems faced by CATV networks is a lack of bandwidth through which to pipe the ever-increasing number of new services that many customers demand. Chen (2001) suggests the

20 Mpeg-4 over CATV embedding of additional content in video transmissions as a way to increase this capacity. Among the advantages of doing this include: The ability to deliver digital services within the same MHz spectrum as video; The ability to continue to use the analog television and set-top box base that is currently in place: "About 500 MHz of spectrum is typically reserved for analog television broadcasts By embedding digital content into existing channels, operators can 'forward migrate' analog infrastructure into the digital age" (Chen, 2001); This results in the getting of about 100 MHz more digital spectrum. Using an information embedding chip set, a cable operator is able to deliver digital video programming within the analog NTSC channels. Figure 2.2 below presents a schematic of how this process would work: Figure 2.2: Information Embedding NTSC Modulator for Delivering Digital Video Services in Analog NTSC Band 9 The difference between this NTSC modulator and the regular one is that this modulator will accept both types of signals-base band NTSC and digital MPEG-2 video, and then embeds the MPEG-2 into the analog signal: Any existing analog or digital set-top box or any existing television receiver can still demodulate the resulting NTSC signal. An information-extracting set-top box equipped with the appropriate decoder chip can demodulate either the NTSC signal or the embedded MPEG-2 9 Chen, 2001.

21 Mpeg-4 over CATV digital video signal, depending on which signal the viewer desires to watch (Chen, 2001). Significantly, this does not require new equipment: any RF/IF tuner that is able to convert a channel in the RF band to IF will work in all three spectrum areas: analog, pure digital, and embedded digital. This also applies for the MPEG-2 decoder which can be used in both pure digital and embedded digital spectrums. 2.4 Summary of Findings MPEG compression techniques, combined with the powerful set of tools available for encoding/decoding digital video, are making MPEG-2 the odds-on favorite to become the de facto standard for the delivery of digital video to home users. As well, because it is possible to transmit MPEG-2 encoded video through analog CATV networks, with the proper encoders and decoders, this will allow present-day cable operators to slowly phase in new equipment without forcing clients to spend increasing amounts on upgrading. As well, the cable operators themselves will not have to lay new fiber optic cable until they are prepared to do so-and still be able to present their customers with the possibility of viewing MPEG- 2 digital video. However, it should be noted here that there is no guarantee presently that this same technology (for example the embedding document material) can be used for the transmission of MPEG-4 digital video through analog CATV networks. The next chapter deals specifically with MPEG-4, how it differs from MPEG-2, and how that might affect the transmission of MPEG-4 digital video-or even if it is theoretically possible to transmit such video over analog CATV networks.

22 Mpeg-4 over CATV MPEG MPEG-4 Compared MPEG-4 is one of several digital video compression systems competing to meet various expected customer expectations in this area. Among the others are RealPlayer from RealNetworks, Windows Media Player from Microsoft, and QuickTime from Apple. The advantage MPEG-4 has, aside from its already established previous iterations MPEG-1 and -2, is that it was created and streamlined to work on consumer-electronics devices (with their less than hefty CPU power and cost) versus the other three which need the micro processing power of PC desktops. According to Ryan Jones, media and entertainment consultant for the Yankee Group: "Consumers want the movie experience in their living room" (Quoted in "Goodbye to the Video Store," 2002). This would mean that the set-top box would evolve into one of the cornerstones for viewing digital video. Another advantage held by MPEG-4 is that it contains a large and robust toolbox, which allows a user to adapt any one or more of its 23 mathematical profiles for any number of devices, from pocket PCs to set-top boxes. Taken all together, the above-cited advantages give a considerable edge to MPEG-4 in the selection from among existing competing technologies. 3.2 MPEG-4's New Paradigm Pereira (n.d.) argues that the MPEG-4 standard allows a move away from what he calls the "Television paradigm," i.e., a two-dimensional view of the world, towards one in which the user can not only watch what is taking place but also interact. In other words, MPEG-4 helps bring together or bind the increasing convergence among the worlds of communications, computing, and TV/movies/entertainment. Therefore, unlike MPEG-2, which operates at high bit rates, MPEG-4 offers:

23 Mpeg-4 over CATV All types of data representation from video (low bit rate to very high quality) and music to 3-D objects and text; The ability to manipulate various objects within a scene; The ability to interact and hyperlink; The provision of a delivery system that is not dependent on the representation format, thus giving it the ability to be used across a wide variety of delivery environments. The basis for this is the new encoding and compression system employed by MPEG-4: object-based rather than frame-by-frame. This means that the scene is created through the use of individual objects using their relationships in space and time rather than entire frames. The advantages of this include: (1) the ability to represent different objects differently when it comes to compression; (2) the ability to integrate various types of data all in one scene (cartoons and real life action, for example); (3) the ability to interact with objects. Figure 3.1 below provides a schematic representation of the MPEG-4 architecture. Figure 3.1: MPEG-4 Object-Based Architecture Pereira, n.d.

24 Mpeg-4 over CATV The MPEG-4 standards consist of six parts including: Systems: scene description, multiplexing, synchronization; Visual: coded representation of both natural and synthetic visual objects; Audio: coded representation of natural and synthetic audio objects; Conformance Testing: conformance conditions for bit streams and devices defined (used to test implementations); Reference Software: software corresponding to both normative and non-normative tools (used to implement compliant products); Delivery Multimedia Integration Framework (DMIF): defines session protocol for management of multimedia streaming over generic delivery systems. MPEG-4 started out as an attempt to simply increase in a marked way the compression ratio that had been achieved in MPEG-2. However, it was soon realized that this could not be accomplished to the extent of justifying a new standard (just like MPEG-3 was finally folded into MPEG-2). The objectives needed to be expanded. This is when the idea of working with convergent technologies came to the fore-to "support new ways, notably content-based, for communication, access and manipulation of digital audiovisual data" (Pereira, n.d.). 3.3 MPEG-4 Tools Aside from the tools inherited from MPEG-1 and MPEG-2 (Systems Target Decoder and Packetization of Streams), MPEG-4 has an entirely new set of tools including: Systems Decoder Model: Because MPEG-4 streams can be different from previous ones, it was necessary to ensure that the way content is transported is not integrated into the architecture;

25 Mpeg-4 over CATV Sync Layer: Used to encode the synchronization information needed to ensure that MPEG-4 can address a bit range from a few kbps to several Mbps; FlexMux (Flexible Multiplex): Used to enhance the transport of MPEG-4 content in an environment where these data streams can behavior unpredictably over time and can occur in very large numbers. 3.4 The MPEG-4 Delivery Layer Avaro et al (n.d.) describe the mission of the MPEG-4 Systems as: "Develop a coded streamable representation for audio-visual objects and their associated time-variant data along with a description of how they are combined." One of the keys to the flexibility of MPEG-4 (and its potential for use as a way to deliver streaming digital video) is the fact that MPEG-4's design is separated from any specifics as to the delivery system being used. This was done through the DMIF (Delivery Multimedia Integration Framework) Application Interface (DAI) which "defines the process of exchanging information between the terminal and the delivery layer in a conceptual way" and "defines procedures for initializing an MPEG-4 session and obtaining access to various elementary stream that are contained in it" (Ibid.). In fact, DAI has a filter which can handle multiple transport requests at the same time. For example, while one DMIF can specify satellite broadcast, another might request IP multicasting-thus MPEG-4 can support "simultaneous transmission of multiple streams over multiple transport technologies and protocols" ("A Guide to MPEG-4," n.d.). Another key element to be considered when it comes to transport issues is the fact that MPEG-4 can be delivered through MPEG-2 Systems. This means, in effect, that (1) MPEG-2 and MPEG-4 data can both reside within an MPEG-2 multiplex; and (2) MPEG-2 data can be referenced within the MPEG-4 scene description:

26 Mpeg-4 over CATV MPEG-4 Logic The list of the various potential applications for MPEG-4 highlights the logic behind separating the delivery system by providing a non-specific delivery layer. MPEG-1 defined how to store data on a file for the purposes of local retrieval; MPEG-2 defined two separate systems: one for local retrieval and one for TV broadcast over established networks. So successful has MPEG-2 become that it is now the transport protocol of choice for delivering digital TV signals on such networks. On the other hand, MPEG-4 "has been targeted since the beginning to adapt to multiple operating scenarios (local retrieval, remote interaction, broadcast or multicast) and delivery technologies. Instead of defining a number of optimized monolithic variations, the design choice was to abstract the functionality that the delivery layer has to provide, and focus the MPEG-4 Systems activity on the common features" (Franceschini,.n.d.). Figure 3.2 below shows how the three layers-compression, Sync, and Delivery-work in MPEG-4: with the Compression Layer doing the encoding and decoding of Elementary Streams; the Sync Layer managing the Elementary Streams as well as their synchronization and hierarchies; and the Delivery Layer ensuring transparent access to the content provided without taking delivery technology into account. Figure 3.2: MPEG-4 Layered Model This means that any information and details related to the delivery technology are kept away from the other two layers, providing an architecture that

27 Mpeg-4 over CATV can handle all three types of delivery technologies: local, interactive remote, and broadcast. Figure 3.3 shows how MPEG-4, through DMIF, manages all three major technologies. Figure 3.3: Addressing the Delivery Integration of Three Major Technologies Aside from separating the compression and sync layers from the delivery layer, DMIF also hides the operational scenario details. The common interface to the delivery system is used to manage access to all streams: locally or remotely retrieved, as well as broadcast/multicast. This means that any differences among the operational scenarios would not affect the interface nor on the application's management of content: For example, MPEG-4 scenes meant to be used in a broadcast environment will not activate features like reverse playing that would be possible instead when playing content designed to work in a local or remote retrieval environment (Franceschini, n.d.). This makes it easier for an application that intends to use different kinds of content as well as content that mixes local and remote retrieval, and/or broadcasting/multicasting.

28 Mpeg-4 over CATV Broadcast Scenario In a broadcast scenario, representation of services is done through a bundle of streams delivered over a set of channels. The request for a particular service by an application leads to a parsing of a specific DMIF URL by the target application module in order to determine which service was requested. As well, the Original Descriptor (OD) for the service and associated Stream Map Table are retrieved. At this point, the target application module can fulfill further originating application requests for the addition of channels. This is done through the comparison of requested ES_ID with the appropriate elements in the Stream Map Table-thus locating the physical channel that carries the stream that was requested. The problem being faced at the moment is that, while static Stream Map Tables are easily handled through extending already existing tools, these tables are actually dynamic: channels can be added or subtracted while a retrieval session is going on. That is why at present the focus is on MPEG-2 TS broadcast, as the Stream Map Table here is shown through an extension of the MPEG-2 Tables. According to Franceschini, while it might seem at first glance that DMIF does not pay any attention to elements that already exist, the opposite is actually taking place: DMIF deliberately chose to specify the very minimum needed to pave the ground to the future exploitation of MPEG-4: it defines a reference architecture to consistently select and make use of the existing tools, but does not force the usage of any specific technology (n.d.). Advantages of using the DMIF on MPEG-4 Systems include: Providing a sharp, clear delivery and operational scenarioindependent mechanism. Ability for systems walkthroughs to be maintained in a unique and general way. Because all walkthroughs start at DAI, the ability to analyze in a consistent manner all delivery technologies.

29 Mpeg-4 over CATV The unambiguous definition of parameters requiring exposure at the DAI. Ensuring that various delivery technologies and operational scenarios can be employed without impacting on the systems features. Providing a formal way to determine how cross-referencing between services can be resolved-through the use of a common syntax in the form of the URL. Such URLs point not to specific content but to a service which keeps the content hidden. 3.7 The IM1 Implementation Software has been created in which the DMIF reference architecture has been validated through implementation of several instances (Franceschini, n.d.). This software has been included as part of the MPEG-4 reference software bundle. The DMIF portion of the software includes the DMIF Filter and the MPEG-4 FlexDemux, and the instance by which streaming content is accessed from local files. As well, other DMIF instances have been developed, including: delivery of MPEG-4 content over the MPEG-2 Transport Stream, IP unicast, and IP multicast. According to Franceschini (n.d.): The interface between the DMIF Filter and the various DMIF instances is realized by means of a couple of classes definitions, one for each direction of the flow. In particular the interface defines a base class from which all DMIF instances shall inherit their own specific derived class. The derived class characterizes the DMIF instance, but the DMIF Filter only invokes the methods defined in the base class. This allows the DMIF Filter to control DMIF instances that are not known in advance, and thus ensures one key feature of the DMIF architecture, that is to allow full independence of an application from the delivery technology used. Among the DMIF instances included in IM1 are: local retrieval scenario; multicast scenario; and remote retrieval scenario. The instances helped to validate the key concept in the DMIF reference architecture-its ability to shield

30 Mpeg-4 over CATV applications from the delivery systems and operational scenarios. Franceschini (n.d.) concludes that: "The DMIF specification defines an architecture that is open to future evolutions in the delivery technology, and that is able, if actually implements in the terminals, to protect investments in the development of multimedia applications." 3.8 Summary As previously indicated, thanks to its levels and profiles, MPEG-4 can be enabled in a variety of applications ranging from large-screen, high-quality video to black-and-white security systems, from desk top streaming video to set-top video on demand. Thus, each individual application can use only the tools it needs to perform the task at hand. For cable operators, for example, it is not presently necessary for an MPEG-4 encoder to be able to take real video from a natural image and break it down into its constituent objects (a very complex procedure) so that an end user can manipulate those objects. Instead, what the cable industry likes about MPEG-4 is that it has a very flexible bit-rate-ranging from very low to very high. The low bit-rate ability would allow the broadcasting of MPEG-4 streams through an MPEG-2 transport. Or it could be delivered point-to-point over an IP network. According to Peter Ausnit, equity analyst at Prudential Volpe Technology Group: The industry today is streaming limited-video-quality clips to PCs. But over time, it's entirely possible for cable operators to deliver almost an unlimited amount of high-quality video streams to TVs, with channels 1 to 300 dedicated to broadcasting digital TV programs and channels 301 to 3 million for streaming high-quality video on demand (Quoted in Yoshida, 2000, p1). However, it is very feasible and conceivable that cable operators would want that type of interactivity further down the road-if only to be able to compete successfully with other modes of transport and delivery. This has spearheaded the conversion of much of the cable infrastructure to two-way hybrid

31 Mpeg-4 over CATV fiber/coax which can stream video through either MPEG-2 transport protocols or IPs. According to Yoshida (2000): Object-based technology would let content providers create personalized video streams for consumers. Advertisers, for example, could change the billboards shown in the background during hockey games or could change the color of a car in a TV commercial according to their needs or a viewer's preference (p1). That MPEG-4 will be the compression choice for such video streaming has yet to be determined. For one thing, decisions need to be made concerning the what the best levels and profiles are for streaming video over cable: "For cable, a standard streaming video profile would ensure that the set-top could detect incoming MPEG-4 objects to assure it could download software and to establish that its CPU could decode a number of objects. The box could then render those objects in software and overlay them on MPEG-2" (Yoshida, 2000, p1). At the moment, MPEG-4 seems a good candidate to perform this task. Pereira (n.d.) concludes with: MPEG-1 and MPEG-2 have been successful standards that have given rise to widely adopted commercial products, such as CD-interactive, digital audio broadcasting, and digital television. However these standards are deeply limited in terms of the functionalities provided by the data representation models used. The recent MPEG-4 standard opens new frontiers in the way users will play with, create, re-use, access and consume audiovisual content. In the next chapter, the paper examines how MPEG-4 came to its present specifications: through the verification models and the core experiments conducted in various parts of the world. This is an ongoing process with experiments still being conducted.