Compression: An evolution, not a revolution

There is optimism that H.264 will mature to compression levels that surpass the impressive levels available today.

The last decade has seen a tremendous increase in the use of video in digital format as broadcasters and broadband operators have attempted to fulfill an insatiable demand for higher volumes of content and reduce transmission/bandwidth cost.

Image placeholder title

The Continuum DVP D9030 MPEG-2 encoder from Scientific-Atlanta supports DVB, PAL and NTSC.

MPEG-2, the present industry standard, was first developed to enable broadcasters to move content from analog to digital, and it started an enormous revolution in content distribution and programming availability by delivering as much as 10 times more programming in the same bandwidth. What is more, it’s still sufficient for most applications today, even after 10 years.

The latest generation of compression technologies such as MPEG-4/H.264 or VC9 are being seen as a way to increase bandwidth efficiency further and provide new services such as HDTV and video on demand (VOD) in new markets, as well as allow telco operators to offer video-over-DSL.

New compression technologies are being developed continually, and from time to time they are incorporated into standards. About 10 years ago, the MPEG-2 standard was finalized and has since been the workhorse of the broadcast and broadband industries ever since. The main driver for developing new compression technologies is the need to reduce the bit rate required to deliver a certain quality of video and audio. This should result in lower transmission cost and/or enable new services to be established — for example, video-over-ADSL.

Even within the current MPEG-2 standard, the development of the actual compression and pre-processing engines has caused significant bit-rate reductions over the last 10 years. Technologies such as statistical multiplexing, pre-analysis and pre-processing now provide the same video quality at about one-third the bit rate needed by MPEG-2 at its outset.

Several new compression technologies have been developed during recent years — many of them being proprietary formats. Generally, they offer bit-rate improvements of approximately a factor of two compared to MPEG-2. The technical community has developed one technology that has been officially standardized called H.264 (or MPEG-4 part 10). The advantages of a standard like H.264 are that it is open and fully backed by the entire industry. Thus, it will likely be continually developed and refined, as was MPEG-2, as engineers from various companies constantly improve its performance over time.

If you consider that H.264 features a much higher number of compression tools and modes than MPEG-2 video and that MPEG-2 video’s compression efficiency improved significantly over the course of 10 years, there is optimism that H.264 will mature to compression levels that surpass the impressive levels available today. An important thing to remember, though, is that many of the MPEG-2 improvements during the last decade related to the advent of statistical multiplexing (variable bit-rate encoding of several channels simultaneously) and advanced pre-filtering and analysis (dual-pass) — rather than improvements in the compression engine itself. H.264 can be directly mapped into MPEG-2 transport streams, enabling the whole existing transmission infrastructure to be reused; only the encoding and decoding devices need to be changed.

H.264 deployment
The big question now is where new compression schemes such as H.264 will be deployed. That is, which applications can benefit from the added compression level offered? In general, this is the case for applications or markets where the bandwidth is limited or expensive for a particular type of service, and where there is no significant population of receivers (STBs) in place for the services in question. Two areas fulfill these requirements and will likely be the first areas where H.264 may be deployed.

The first application is video-over-xDSL systems, especially ADSL. In ADSL networks, the access bandwidth is severely limited for video applications, and there are virtually no receivers yet deployed. H.264 or VC9 could permit a more cost-efficient deployment of video services in addition to Internet and phone access. This, of course, requires the new receivers to become as cost competitive as current MPEG-2 receivers, because new access technologies such as ADSL 2, ADSL 2+ and VDSL in turn increase the throughput bit rate. Thus, it requires some further analysis for each system before the optimum deployment strategy can be determined.

Another large application area is HDTV. HDTV is currently being deployed in Australia, Japan and North America, and it is already putting some constraints on the networks. The latest concept of HDTV-on-Demand (HDOD) requires even more bandwidth. As no HDOD STBs are deployed yet, we might see a new type of STB supporting both MPEG-2 based SDTV and HDTV as well as H.264-based HDTV for on-demand services. The industry is working towards the availability of such combined STBs, co-existing with deployed MPEG-2 STBs in the network to support this application.

Furthermore, enhancements in network throughput by Gigabit Ethernet technology and switched broadcast services can also free up bandwidth in the transmission network. In the USA, H.264-based HDTV could also be envisioned in geographical areas where HDTV is not deployed yet, or in new systems such as Cablevision’s VOOM satellite HDTV service that plans to switch to new compression schemes within the next two years.

When HDTV arrives in Europe and Asia, it will probably be driven by commercial Pay-TV operators that today are mainly found in the satellite (direct-to-home) space. Because transponder capacity is expensive and no HDTV STBs are installed at all in Europe and Asia yet (except Australia), the use of H.264-based HDTV would be a perfect match.

The timing would also seem to be about right given that it will still take a few years before HDTV is widely deployed in Europe. The STBs would probably be dual decoders supporting both MPEG-2 SDTV and H.264 HDTV. HDTV for cable would probably follow quickly after satellite in Europe. Even if transmission capacity is less of an issue in European and Asian cable networks (compared to satellite and North American cable networks), H.264 could probably be used for HDTV from the beginning, as it will certainly be deployed in other systems such as satellite.

On the other hand, it is obvious that many applications and markets will continue to use MPEG-2 for many years to come. For example, existing satellite, cable and DVB-T systems offering SDTV will continue using MPEG-2 because the number of installed receivers is already quite high. And for new systems, the cost of an MPEG-2 STB is still considerably lower than prices projected for H.264, making it more economical to deploy the proven and lower cost technology. This means that MPEG-2 performance can be expected to increase, albeit at a much less dramatic pace.

For systems with existing MPEG-2 based systems, several different migration strategies could be envisioned. Most likely, a service-based approach is most attractive, as new receivers are usually needed for a new service anyway. This also means that purchasing new encoders for MPEG-2 applications that can be upgraded to the new compression technologies in the future may not be the best choice for operators. First of all, these devices initially will be more expensive because the new compression scheme requires significantly more processing power than MPEG-2. Secondly, because the cost of new compression encoders will decrease over time and the encoding performance will increase, waiting may be a wiser choice. Thus, if MPEG-2 is required now, it’s probably best to purchase the best and most cost-efficient MPEG-2 solution available today, and later when H.264 is required, for new applications, new encoders can be evaluated and purchased at that time.

Going forward, both MPEG-2 and other compression schemes will be important in delivering content to subscribers. Some applications can benefit from the new compression technologies like H.264 now. Other applications will continue to use MPEG-2 for many years and then switch to H.264 at a later stage in an evolutionary, rather than revolutionary, approach.

Bernd Reul is the director of strategic marketing and business development for Scientific-Atlanta in EMEA.