Digital TV
A Technical FAQ
Rev 4.0.  June 14th 2000

by Chris Muriel
chris.muriel@analog.com

What is Digital Satellite Television
Part 1) What is MPEG ?

MPEG stands for Moving (or Motion) Picture Experts Group & is an organisation of interested parties. It's run in similar manner to JPEG (Joint Picture Expert Group) -JPEG being for still images; there is also a standard known as M-JPEG (Moving JPEG) but this is intended more for the needs of the security industry. How about the fact that discservers, and corresponding high-end editing suites are only just moving from Motion Jpeg to MPEG (4:2:2). As an intra coding only system it has been and still is widely used in these types of video applications.

MPEG describes a form of compression for digital data where the data represents moving images of a TV like-nature. The standard also allows for audio datastreams sync'd with the video. MPEG1 is common on IBM PC's (& other platforms) using *.mpg files. Xing, Mediamatics & others supply software players for these and all but the cheapest PC VGA cards seem to have some hardware support for MPEG1 files; normally you need a Pentium PC to have much chance of playing mpeg1 files at reasonable speeds (25 frames per second or more).

Anyway, mpeg1 isn't used for satellite TV; the industry needed a faster, more flexible & efficient method. For broadcast use, less tendency to pixellation or "blockiness" was desired with fewer "artefacts" -technical/marketing term for unwanted material on the screen (that's a bit like calling a software bug an "anomaly" !!). Now what the satellite industry wanted was to squeeze more channels into the bandwidth taken up by a satellite transponder. Analogue satellite TV uses around 27-36 MHz of bandwidth for its FM video + audio FM subcarriers; this is for each channel. So the operators want to put 5,10 or more separate channels, via a digital datastream into a similar bandwidth. This allows many more channels or needs fewer transponders to transmit a given number of channels. To give flexibility,the actual compression ratio can be varied between "Studio Quality" and "Video recorder quality".

Studio needs 12 MBits/second data,
broadcast needs 8 MBits,
VHS needs 2 MBits/s.

I don't want this faq to become too technical but read DVBFAQ.TXT from Markus Kuhn for technical info. Just remember that the compression ratio can be varied to cope with the needs of the supplier of the video information. The digital data from several channels can be multiplexed into an MPEG Transport Stream, along with various (compressed) audio channels (which can include digital surround sound & multiple languages). A package with several channels modulated onto 1 carrier is often referred to as a "bouquet" which is, I guess, a marketing term for such a multiplex or package. The compression ratio used can be different for different channels within the multiplex and can vary over time. The lowest compression ratio will occupy the highest bandwidth but will be needed when the video content requires much detail with lots of changes & movement occurring.

A good example might be an athletics event; you would have people running around the track , people throwing javelins and others just walking slowly or sitting down. Consequently there will not be a lot of redundancy in the whole scene so if too much compression is applied, some movement could easily be displayed jerkily. A studio news broadcast, when they are simply showing the newsreader and his desk, will compress very efficiently & occupy a relatively small bandwidth. So the newscast will be allocated a lower bit-rate & compressed more. The whole process of dynamically allocating bandwidth per channel in a multiplex is known as Statistical Multiplexing & uses quite advanced mathematical algorithms to do its allocation processing.
Incidentally MPEG1.5 is a hybrid (falling between 1 & 2) but any mpeg2 receiver SHOULD be backwards compatible with older MPEG revisions.

Actually MPEG1.5 isn't a real (ratified) standard but covers several proprietary systems which tend to improve on MPEG1 . Examples include "System2000" from NTL & "Orbit" from Scientific Atlanta. 1 interesting point is that you can't compress "noise" with current MPEG schemes. Imagine watching a film via digital (mpeg) satellite TV where, within the plot of the film, the camera zooms in to show a TV screen switched on but with no antenna connected. You expect to see nothing but noise ("snow)" on the picture. This signal is entirely random & so can't be compressed - there's no repeating pattern/redundancy in the signal. Apparently a future MPEG version (mpeg4 ?) will have some kind of algorithm built in to get around this problem. It looks like the next "official" (ratified) standard will be MPEG4. Mpeg2 is also used for DVD (digital video disk or digital versatile disc) & other digital video delivery systems including cable,fibre & digital terrestrial TV.

MPEG4, when it comes, is a scalable encoding/compression system so that mini versions could output to the tiny screens on mobile phones and PDAs whilst higher resolutions & larger object sizes would output to televisions. It's more flexible than MPEG2 to allow this object-based scalability as well as interactivity where desired. It introduces its own new acronyms like AVO (Audio/Visual Object) & DMIF (Delivery Multimedia Integration Framework). MPEG4 allows for 2D & 3D images & user selection (or ungrouping) of parts of the whole scene - where the originator has permitted this. It will require more powerful processors to implement these procedures in satellite receivers.

The picture below attempts to illustrate the concept.