Bob's Byte: Deja Vu
By Bob Emmerson, Thu Aug 17 00:00:00 GMT 2000

Don't expect high speeds from regular GPRS. Networks will stay in first gear for a while, devices being the bottleneck. However, compression software can turbo-charge downloads and take us into EDGE territory.

Let's recall the hype. GPRS will deliver packet-switched data at 115 Kbps. Great. It's an always online connection, so e-mail comes through automatically. Terrific. When is GPRS coming? In time for Christmas, but Santa might not be able to deliver the devices. And if he (or she) does, they will only enable access at about 20 Kbps. So what went wrong?

In a word: nothing. Exaggerated claims are an integral part of today's high-tech economy. If all eight time slots were employed you would get 115 Kbps, but the first devices will only have one or two slots. These will be followed by products having three or four slots, which will lift performance to 56 Kbps.

That figure is surprisingly precise: an engineer would say "somewhere between 50 and 60 K". But marketing types know that V.90 modems max out at 56K; it's upper limit for an analogue, wireline connection, so why should wireless be expected to deliver more than wireline? That's fast, isn't it? And remember that you're always online. What more do you want?

Well, call me old fashioned, but a more honest approach would have been nice.

Reality Check

It now looks as if services will start at around 20 K. At first sight this figure is somewhat disappointing, but ask yourself what apps will benefit most from a packet switched service and what kind of device you will be using. The most likely answers are: (a) e-mail and information services and (b) a WAP-compliant phone or a wireless-enabled PDA. These are lightweight apps running on thin-client devices, so it does not matter if the data packets trickle through. The transfer rate is not particularly relevant as long as the price is right. Thus, bits per second do not matter too much right now: what is important is to get the wireless data revolution rolling.

And once the revolution is rolling, the market will want higher rates and this involves buying new three or four slot devices. Not good news. And circuit-switched data could also do with a boost since it will take a while before GPRS roaming agreements are in place. So, is the roll going to be slow and disappointing, or will compression technology come to the rescue?

We've Been Here Before

Nortel has a product that cranks up the rate and the performance figures look pretty good, but we've been here before. A standard called V.42bis was going to enable wireless transfer at ISDN-type speeds (64 Kbps). It was approved by ETSI but the network operators never implemented the technology. That was back in 1996! This time, however, there are two important differences.

One, this is a client/server product so there is no network dependency. There is a Windows client and a proxy server component that resides on a corporate intranet.

And two, Nortel's compression technology uses file-specific data compression and optimisation. The proposed ETSI standard was generic, i.e. all file types would have been compressed in the same way.

The company claims that this will enable Internet access and other data services on 2G (9.6 Kbps) networks to run at wireline modem rates or even higher - it depends on the bloat. PowerPoint files are very bloated and compression can go up very much higher, e.g. x25. However, users still have to live with 30-40 second dial-up times.

Nortel tells me that the product can be used on packet-switched networks, but the relationship is not linear and they are also keen to emphasise that first-gen devices introduce a bottleneck.

Typical improvements are 2-10 times for Web browsing, while MS Exchange e-mail with attachments can improve by a factor of 4-15. Therefore, if an app is running over GPRS at 20K, then adding the compression algorithms could increase throughput to 200K. And when better devices come to market and networks start running at 50 K, the rate could even exceed that of EDGE.

Deja Vu With a Difference

Operators do not have a good track record on the mobile data front: no, let's be honest, their record is abysmal. Failure to adopt the V.42bis standard is just a single, small example. The big difference this time is the ability to run packet-switched, client-server applications, i.e. now we have mobile IP.

IT managers can therefore implement client-server compression whenever and wherever they want. Wherever refers to the need to enable the same service in all the countries in which the company operates. The various networks are simply used as transport mechanisms. And if rates can be pushed up to EDGE levels, then one has to question the business case for this technology.

Right now we have WAP. Soon the market will be asked to trade up to WAP over 2-slot GPRS. Then we get devices with more slots and then finally there's EDGE. Actually, this is not the final stage since 3G is on the radar screen.

I can't buy into this scenario. I don't see the market trading up every 6 to 9 months. And the rebuttal that operators will subsidise the new devices is shallow because that means higher tariffs.

So what will happen? I don't know and I don't think anybody really knows. That's what makes writing about this industry so interesting and exciting.

Bob's Byte is a regular column on TheFeature. Bob Emmerson observes and writes on the wireless industry from his home in The Netherlands.