The Internet Protocol Journal - Volume 1, No. 3

Digital Television: A New Venue for the Internet

by George Abe, Cisco Systems

The digitization of television is of interest to the Internet community in that it opens the possibility of a new mode of delivering IP packets to the home. IP services can be delivered over television broadcast distribution networks, whether over the air, cable, or satellite. This article introduces the basic concepts of digital television (DTV) and provides a point of departure for further reading.

Why Is Digital TV Happening?
The original motivation for the research into advanced TV (we avoid the term DTV for a moment) was to prop up sagging TV sales. It was mostly vendor push.

By the late 1970s, Japan and Korea had achieved domination in the production of TV sets worldwide. They were so successful that the market had become saturated, particularly in the developed world. Everyone had one or, more likely, three or four TVs at home. Further, a TV lasts over 10 years, so the replacement market is low. TV production had ceased to be a growth market. Margins were and are poor and few innovations were on the horizon.

So in the early 1980s Japan had begun research into new high-definition televisions that would stimulate new demand and enable them to keep their market leadership. Their system is called Multiple Subnyquist (MUSE). MUSE was an analog system, but it had better-quality pictures.

Not to be outdone, the U.S. decided it needed to try to recapture the TV market, so began its own development, under the aegis of the Federal Government. A partnership called the Grand Alliance was formed, and it began working in 1984. Pioneering work was done by the partnership members, particularly Zenith, MIT, and General Instruments. They created a digital specification after more than a decade of research and development. Along the way, the computer industry made contributions (or some would say interferences) of its own until the FCC announced a final specification in December 1996. The basic elements are found at www.atsc.org and referenced later in this article.

Benefits of DTV
The movement toward widespread DTV gained momentum among government officials, broadcasters, and hardware vendors when some of the benefits became clear.

First, because of improvements in technology, it is possible to transmit pictures and sound of significantly higher quality in the same 6 MHz spectrum that analog TV occupies. The 6 MHz spectrum is wasteful of bandwidth, and the government would like to recover the excess so it can be auctioned or used to support other public services (police, fire, deep space probes, and so on), which could operate at the relatively low frequencies of VHF TV.

Second, digitally encoded TV could provide new services, such as Web access via TV or interactive TV. These have long been dreams of the consumer electronics (CE) industry, but hope springs eternal. Third, digital TV offers greater security to the programmer and the network. There is a cottage industry in hacking analog set-top boxes. Digital techniques, such as the Data Encryption Standard (DES), double DES, and triple DES give operators hope that they can secure their pay-per-view content.

Finally and most interestingly, since digital TV occupies less bandwidth per program, broadcasters, satellite operators, and cable operators have the opportunity to offer more channels. Instead of a mere 10-13 channels available over the air in a single metropolitan area, it is possible to have perhaps 60 or more over the air channels. Cable operators, with their greater bandwidth underground, could have many more channels. Although technically cable could offer 500 channels, it is hard to imagine where the scripts would come from.

What Is DTV?
By our definition, digital television is the capture, production, distribution, and broadcast of programming in a digitally encoded format. Whereas today's analog TV transmits in amplitude modulation, DTV would use Quadrature Phase Shift Keying (QPSK), Quadrature Amplitude Modulation (QAM), or Vestigal Side Band (VSB) modulation techniques. We won't detail these techniques here except to mention that they are mutually incompatible.

When DTV standards were discussed in the 1980s, the industry could not agree on a single display. The deliberations became more protracted with the entry of the computer industry into the discussions, long after the broadcasters and consumer electronics people began their work. Would there be interlaced or progressive scanning? Would there be the existing aspect ratio or would there be a widescreen display? Square pixels or not? How many lines of resolution would be displayed?

With the broadcasters and consumer electronics vendors arguing for interlacing, oval pixels, and wide screens and the computer people arguing for progressive scanning, square pixels, and a more square display, the disagreements could not be bridged.

Therefore, the FCC had no choice but to declare that the "market should decide" which display format would prevail. Accordingly, the FCC announced in December 1996 that 18 different display formats would be permissible for over-the-air digital TV. A broadcaster could elect to transmit in any of the approved formats. The approved formats are shown in Tables 1 and 2.

Table 1: Progressive Video Scanning Formats for Digital TV
Vertical Lines Horizontal Pixels Aspect Ratio Frame Rate per Second
1080 1920 16:9 24, 30
720 1280 16:9 24, 30, 60
480 704 16:9 24, 30, 60
480 704 4:3 24, 30, 60
480 640 4:3 24, 30, 60

Table 2: Interlaced Video Scanning Formats for Digital TV
Vertical Lines Horizontal Pixels Aspect Ratio Frame Rate per Second
1080 1920 16:9 30
480 704 16:9 30
480 704 4:3 30
480 640 4:3 30

The vernacular to describe the formats typically indicates the number of vertical lines and the scanning format. For example, "1080i" refers to 1080 lines, interlaced scanning; "720p" refers to 720 lines in progressive format.

In practice, only a few of the 18 approved formats are under consideration by the nation's broadcasters. NBC and CBS have declared they will support 1080i. ABC is opting for 720p, and Fox has opted for 480p.

Apart from the controversy over display, most of the other elements were quickly resolved. Modulation scheme, transport multiplexing, compression, timing, and an overall systems and testing procedure were agreed to. The apparatus for DTV was in place, almost. The time was January 1997.

High Definition or Standard Definition
Some view DTV as synonymous with high-definition television. It is not. DTV encompasses both High-Definition TV (HDTV) and Standard-Definition TV (SDTV). Hence HDTV is a proper subset of DTV. The difference between HD and SDTV is not standardized, but our definition of HD includes the display formats that have 720 or 1080 lines. Formats with fewer lines are standard definition.

The key point of difference between HD and SD is that with HD and current compression techniques (MPEG-2), only one program is accommodated in one 6-MHz channel. With SD, it is possible for the broadcaster to transmit two or more programs simultaneously, in a single 6-MHz chunk of bandwidth.

This has tremendous implications. If broadcasters can transmit multiple channels at once, it would be possible (technically) for Disney to broadcast ABC, the Disney Channel, ESPN, and A&E over the air in the same bandwidth they use to show ABC today. (Of course they won't do this for commercial and contractual reasons, but the technology makes it doable).

For Internet Service Providers, a broadcast could transmit SD programming simultaneously with datacasting, and go into the push-mode data service business. For example, Disney/ABC could download software updates for Disney Interactive, or perhaps contract with Microsoft to deliver Windows updates. Whereas most Internet folk view MPEG being transported inside IP packets on the Internet, broadcasters intend to insert IP packets into MPEG-2 transport streams. The consumer's digital set-top box would tune to the data "channel," extract the data from its MPEG capsule, and divert the data packet to an Ethernet or ATM port on the set-top.

There are nearly 1,600 broadcasters in the U.S. Each could, in theory, transmit 19.3 megabits per second. Of course, most of these bits will be used for television, but certainly 1 or 2 megabits can be accommodated by each broadcaster for data service.

Given the dearth of programming to fill multiple SD channels, broadcasters are strongly motivated to consider data services and compete for a slice of the Internet service market.

Digital TV—End to End
Whereas one easily thinks of DTV as a distribution and display technology, in fact there are major changes required to capture, edit, and distribute digital content. Thus there is the need for new cameras, post-production editors, sound mixers, and the like. Digital TV can be transmitted over the air, through cable networks, or via Direct Broadcast Satellite (DBS). Today, only DBS has achieved large-scale distribution of digital TV, with over 7 million subscribers in the U.S. and 15 million worldwide.

Content is created either through a digital camera or by converting existing analog content, such as 35mm film, into digital format. Within the production environment, editing changes are made, typically using Nonlinear Editors (NLEs) that connect to a local-area network.

Original production is normally done in the high definition. The highest form of resolution is 1.492 Gbps. (See Table 3.) Equipment to do this is not widely available, but it will be eventually. Panasonic is shipping a digital camera capable of 1.5-Gbps output, but rumor has it they cost almost $500,000, if you can even get one. Nonetheless, 41 stations began HD programming in November, highlighted by an NFL game on CBS between the Buffalo Bills and the New York Jets on November 8.

Some compression is applied within the postproduction and editing environment. The TV industry, through the Society of Motion Picture and TV Engineers (www.smpte.org), developed a series of digital transmission standards. Chief among these is SMPTE 305M, which defines a protocol called Serial Data Transport Interface (SDTI), which calls for a 270- or 360-Mbps service to link various pieces of production equipment such as NLEs in a postproduction facility. SMPTE 305M is a networking scheme complete with an addressing specification. (Interesting point about 305M: It is the first and only protocol known to this author that specifies use of IPv6 addressing.) Another important protocol is SMPTE 259M, which is a link-layer protocol underneath 305M.

A competing protocol to SDTI is the Digital Video Broadcasters Asynchronous Serial Interface (DVB-ASI). Information on DVB-ASI is found at www.dvb.org.

From the editing environment, content is distributed via satellite or land lines to local affiliates (for local over-the-air broadcast), cable head-ends (for cable TV distribution) and satellite hubs (for direct-to-home satellite service). The distribution from national feeds to local facilities is normally at T3/E3 speeds because of the availability of T3/E3 services by telephone companies and satellite transponders for affiliate and direct-to-home distribution.

Cable providers, local broadcasters, and satellite services add their own content and make certain changes to the national feeds. Among these changes are assignment of the programming to specific frequencies or channels, insertion of local advertising, local programming, and emergency broadcasts.

After adding their own content, the local services distribute the final programming to consumers. Over-the-air broadcasters will transmit 19.3 Mbps per 6 MHz, cable will transmit 27 Mbps per 6 MHz, and satellite uses variable channelization, kept closely under wraps.

So there is the progression downward from 1492 Mbps of original encoding, to 270 Mbps for editing, to 34/45 Mbps for affiliate distribution, to 27 Mbps or less for distribution to the end user.

Table 3: Bit Rate Requirements for Various Display Formats
Format Pixels per Line Lines per Frame Pixels per Frame Frames per Second Millions of Pixels per Second Bits per Pixel Mbps
SVGA 800 600 480,000 72 34.6 8 276.5
NTSC 640 480 307,200 30 9.2 24 221.2
PAL 580 575 333,500 50 16.7 24 400.2
SECAM 580 575 333,500 50 16.7 24 400.2
HDTV 1920 1080 2,073,600 30 62.2 24 1492.8
Film 2000 1700 3,400,000 24 81.6 32 2611.2
Note: Film display formats vary, depending on content and directorial prerogative.

Over the Air and Cable
All the huffing and puffing by the FCC, the consumer electronics industry, the computer industry, and the broadcasters pertains to over-the-air transmission. However, about two-thirds of the American viewing public views TV through cable. So if most Americans are to receive DTV, they must receive it through cable.

This raises important technical and regulatory questions. The technical question is: How are the digital signals produced by the broadcasters and their affiliates to be sent through wires, and what is the allocation of functions between the digital set-top and the digital receiver? This question seems simple but it is not, as we shall see.

The regulatory question pertains to whether the cable operators are to be compelled to carry DTV from broadcasters. This problem is referred to as the digital Must Carry Problem, now under consideration by the FCC. It certainly will be litigated, whatever the outcome of the FCC's decision.

Technical Question
Among the key provisions agreed to by the Grand Alliance is the use of a modulation technique called 8-VSB for over-the-air digital transmission. The particulars of 8-VSB are not significant here, but we will mention that this particular decision was arrived at in the mid-1980s, before the cable industry had much impact on the viewing public or on the broadcasting industry.

When the cable industry began to think about digital, in the mid-1990s, they settled on a modulation scheme called 64 QAM. 64 QAM is able to produce 27 Mbps in 6 MHz, whereas 8-VSB produces about 19.3 Mbps. The difference occurs because over-the-air broadcasting requires a more robust encoding scheme to combat the more hostile nature of over-the-air transmission, as opposed to the safer environment of coaxial cables. Thus the cable modulation technique can be more aggressive than over-the-air techniques.

(We should add that satellites use an even more robust modulation technique called QPSK, which gets fewer bits per Hertz than VSB or QAM. But robustness is needed because satellite signals must travel far greater distances than cable or local broadcast.)

Thus for cable to carry a digital over-the-air broadcast, some conversion of 8-VSB encoding to 64 QAM encoding is necessary. This necessity does not present a major technical problem, but agreement is needed on where the conversion is done and at what cost. For example, Broadcom and Sony are collaborating on the development of a chip, to be embedded in a TV, that can decode VSB and QAM. It sounds simple, but the cable industry is not interested. They want to carry QAM and QAM only on their networks.

One option is to convert the format of the digital bitstream coming out of the cable box to the IEEE 1394 FireWire format. Since DTVs are likely to have FireWire input, this conversion can provide a ubiquitous connection. However, this scenario raises the problem of copy protection, a sore point in Hollywood. Since digital copies are pristine, the content providers (studios and record companies) are firm in their resolve that unless there is strong copy protection, none of their content will be available over FireWire.

Another option is to build a set-top box that takes baseband signals and modulates them to look like 8-VSB broadcast signals on channel 3, similar to how VCRs work in the analog world now. This scenario is clearly rather ugly, but understood by consumers.

Finally, it could be up to the cable operators to transmodulate the 8-VSB into QAM at the cable head-end. Better yet, they can accept broadcasters' feeds in baseband, and then QAM-modulate the baseband signals for their consumers. The cable set-top box would be sending bit maps to a dumb digital monitor, like a computer monitor, which doesn't know or care that it is receiving QAM or VSB programming. Apart from modulation, there is the issue of display format. NBC and CBS have declared they will transmit in 1080i. ABC has chosen 720p and Fox has chosen 480p, with some vague pledge for higher definition later. After all, it does not seem necessary to show The Simpsons in HD.

On the other hand, John Malone, Chairman of TCI, went public in May 1998 with his declaration that TCI would not voluntarily carry 1080i because it (1080i) was wasteful of bandwidth. Implied in his comment is the fact that cable operators do need to be restricted to 6- MHz channelization for digital. In fact, the entire DTV spectrum on cable could be considered a gigantic pool of bandwidth that the cable operator could allocate to individual channels, much as direct satellite does. This setup gives the cable operators incentive to downconvert the broadcasters' DTV signals. For example, when NBC sends 1080i, the cable operator may elect to transmit 720p, or less, to its customers.

Should the cable operators be required to carry the HDTV pictures from the broadcasters in the broadcasters' chosen format? Would they be allowed to downconvert the HD into standard definition? What happens when a broadcaster, say NBC, elects to transmit in SDTV and thereby has the capability of multiplexing several channels onto a single chunk of 6 MHz? What is the duty of the cable operator to carry Internet datacasting offered by the broadcasters over the cable network, in competition with services such as @Home and Roadrunner?

The complexities of multiplexing go further. Let's say ABC elects to broadcast SD. If one of the subprograms in the multiplex is a pay-per-view channel, should the authentication procedures of the cable operator be superceded? Should the electronic program guide of the cable operator be superceded?

Questions like these have technical and regulatory aspects and are being worked in industry, the FCC, and state regulatory agencies. It is possible that Congress will get involved as well. When John Malone made his statement, both sides of the aisle in Congress were not amused. They want DTV to happen so that spectrum can be freed. If the cable operators stand in the way, the conversion to digital is stopped dead in its tracks.

The Open Cable Initiative
The cable industry does not want to be a bottleneck to broadcasters. On the other hand, it needs to make quick progress into DTV to compete against satellite. Therefore, the industry has embarked on a process called Open Cable, which seeks to define a digital set-top box that can be available at retail. Available at retail means a nonproprietary, open design. Open Cable strives to make the DTV set-top box independent of processor platform (that is, not an Intel Pentium necessarily) and operating system independent (that is, not a Microsoft Windows CE necessarily).

The Open Cable set-top box will allow for data services through a specification written by the Digital Audio Visual Council (DAVIC—www.davic.org) and therefore, is not compatible with the current Data-over-Cable Service Interface (DOCSIS) specification supported by the U.S. cable industry. (See article starting on page 13.) However, it is possible for DOCSIS capabilities to be added on to an Open Cable set-top box. We mention Open Cable because it will be the key customer premises device for cable and digital TV and much hinges on its interoperability with broadcasters transmissions.

Digital TV via Satellite
In addition to over-the-air and cable, DTV can be received by satellite. As of this writing, it is the only way to receive DTV. The digital satellite industry has nearly 7 million subscribers who received DTV today. Its role in all the discussions of HD vs. SD and the provision of data services is relatively low key because it is believed that satellite will continue to be a niche provider because of its technical and legal problems in distributing locally originated TV stations.

But satellites bear watching because if they are able to deliver local channels and obtain 15-20 million homes in the U.S., then the financial consequences on cable and over the air could be crucial.

The New Digital Studio
The figure shows a schematic of the elements of a DTV broadcast studio described recently by the U.S. National Institute of Standard and Technology (NIST). At the heart of the studio is an ATM switch with new interfaces that connect to DVB or ATSC infrastructures via DVB-ASI or SDTI interfaces.

Connection for wide-area distribution will likely be over ATM. Converters exist for DVB-ASI to ATM. For example, Cellware (www.cellware.de) in Germany markets such a converter, but there is no SDTI-to-ATM interface known to this author at this time. The digital studio provides a new a marketing opportunity for the LAN industry. Broadcast digital production demands higher speeds than most other LAN applications.

Thus vendors of data communications equipment have two opportunities: to provide equipment to broadcasters who want to enter the Internet service business and to production houses that use ATM or other LANs to support editing and production applications.

Figure 1: Prototype of HDTV Broadcast Studio

(Click on image to enlarge.)

Web Sites
www.atsc.org: Advanced TV Standards Committee. S13 and S16 are subgroups working on datacasting; S13 focuses primarily on the downstream path, whereas S16 focuses primarily on the reverse communication from the receiver. Since over-the-air is one way, this work is limited to the communications between the S13 forward channels and a telephone or Internet return path.

www.dvb.org: The Digital Video Broadcasting Project (DVB) has taken the lead in defining DTV specifications as well as defining datacasting interfaces over DTV infrastructures.

www.dvb.org: The Digital Video Broadcasting Project (DVB) has taken the lead in defining DTV specifications as well as defining datacasting interfaces over DTV infrastructures.

www.smpte.org: Society of Motion Picture and Television Engineers.

www.sbe.org: Society of Broadcast Engineers.

www.scte.org: Society of Cable TV Engineers.

www.mpeg.org: Motion Picture Experts Group. The word on MPEG compression, controls, and transmission.

References
[1] ISO/IEC IS 13818-1, International Standard, MPEG-2 Systems.

[2] ISO/IEC IS 13818-2, International Standard, MPEG-2 Video.

[3] ISO/IEC 13818-6, International Standard, Digital Storage Media Command and Control (DSM-CC).

[4] ATSC Standard A/52 (1995), Digital Audio Compression (AC-3).

[5] ATSC Standard A/53 (1995), ATSC Digital Television Standard.

[6] ATSC Standard A/55 (1996), Program Guide for Digital Television.

[7] ATSC Standard A/56 (1996), System Information for Digital Television.

[8] ATSC Standard A/57 (1996), Program/Episode/Version Identification.

[9] ATSC Standard A/63 (1997), Standard for coding 25/50-Hz Video.

[10] ATSC Standard A/64 (1997), Transmission Measurement and Compliance For Digital Television.

[11] ATSC Standard A/65 (1998), Program and System Information Protocol for Terrestrial Broadcast and Cable.

[12] ATSC T3/S13 Doc. 010 DVS-yyy Rev z Draft, ATSC Data Broadcast Specification for Terrestrial Broadcast and Cable.

[13] ETR XXX: Digital Video Broadcasting (DVB); Guidelines for the Use of the DVB Specification: Network Independent Protocols for Interactive Services (ETS 300 802).

[14] SCTE DVS-nn: SCTE Digital Video Subcommittee (DVS) standard for Cable Headend and Distribution Systems (spec not released— under development)

GEORGE ABE holds an A.B. in Mathematics and an M.S. in Operations Research from UCLA. He currently is a Consulting Engineer at Cisco Systems, where he has dabbled in various areas of residential broadband networking since 1994. He is the author of Residential Broadband, Cisco Press (imprint of Macmillan Press). He expects to be a early adopter of digital TV and, when not watching TV, he can be reached at georgea@acm.org