Data is ripe for squeezing

A new data compression standard from the ITU could cut costs and speed up many applications, so it will not be long before almost everyone is using it, says Bill Pechey

Next month, the ITU will complete a new standard that has the potential to reduce the data communication costs of almost everyone. It's a data compression algorithm called V.44.

Previous data compression algorithms have not made much difference except, perhaps, in dial-up modems, but V.44 is different. First, it does a better job of compression than previous algorithms, such as the popular V.42 Bis, with a typical improvement of about 30 percent, and sometimes much more.

Second, it's been designed so that it can be used in almost any data communication system, not just in modems. V.42 Bis requires the underlying data channel to be error-free ­ V.44 does not.

Also, the algorithm's performance improves when given more memory. In contrast, V.42 Bis has an optimum amount of memory for each type of data, beyond which it gets slower.

Lastly, and perhaps most importantly, V.44 adapts very quickly to the type of data being processed. This means that it can compress the contents of a data packet without needing to use the information from previous packets ­ in effect, it starts from scratch with each packet and still achieves worthwhile compression. This solves a problem that has prevented data compression from being widely used in IP networks. Earlier schemes required a network node to maintain a separate compressor for each stream of packets that it had to deal with. V.44 uses just one compressor, which operates on each packet in turn. The saving becomes very significant when you consider that a single browser session might have about eight separate simultaneous connections to a Web site, each carrying a different type of data.

So, where will V.44 be used? The new V.92 modems should be out by Christmas and will be the first large-scale application of the technology. They will be capable of over 200kbit/s in each direction, on a good day.

V.44 is being considered by the mobile phone standards bodies to improve the performance of GPRS and 3G data services. It's likely to take a year or so before V.44 starts to appear in mobiles but its effect will be significant. It will soon appear in routers and similar network devices, which will improve the throughput of the data circuits that connect the routers. Taking all these together, it won't be long before almost everyone will be using V.44, often without realising it.

The V.44 algorithm is largely the work of one man, Jeff Heath of Hughes Network Systems. Jeff is not a compression expert ­ he just thought he could make a better algorithm, and he did.

One of the reasons that V.42 Bis did not spread very far beyond modems was because of the complexity of the patent situation and the attitude of some of the patent owners. So far, only Hughes has declared that it holds patents in V.44, and its licensing policy appears to be very reasonable.

Finally, I have to declare an interest ­ but not a financial one. I'm the chairman ­ rapporteur in ITU-speak ­ of the group that turned Jeff's scheme into a standard.





Home page   Clients   Bill's ITWeek Column   Industry Organisations   Website Design   Staff