Lay it on the line

DSL, ADSL, ISDN - just what do they all mean? Find out why digital subscriber lines are the next big thing in phone technology.

It's the Next Big Thing, far bigger than the previous Big Thing -l subscriber lines are the next big thing in phone technology. whatever that was. It's going to solve all our problems and it will make the internet even bigger than we said it would be.

It is digital subscriber line (DSL) and it will deliver multi-megabits per second to your phone socket as well as making everyone's life better and more fulfilled, or so it says on the press release. It is also not available yet, but this should not worry you, for as we all know, a minor hiccup like availability isn't going to stop the Next Big Thing.

We have been through many iterations of 'get info down the phone line faster' technology. And each was declared impossible by the experts. After all, modems had reached the inherent limit of analogue lines. Then, when another, faster, modem arrived the 'you cannot go faster than (fill in the speed here)' comments were quietly forgotten and a new round of 'you can't go faster than this' started up.

Anyway, users' connections are invariably at lower speeds than the modem is rated. We now have 56K modems, which rarely give a true 56Kbps because of phone line imperfections, and that is only when receiving data, not sending it.

After that is ISDN, which, while helping a bit, is expensive and still doesn't move data fast enough to make a real difference - and its takeup by the rest of the world (that is, the US) is rather limited. Basic ISDN transfers data at 64Kbps and even the next level ISDN reaches a ceiling of 128 Kbps.

The up side is that those speeds are for both sending and receiving, but at best it is still only four times faster than a 33.6Kbps modem, and while they have become less expensive they're not cheap by any means.

The pressure is on. There has to be a new way of getting data across the telephone network. The traditional problem has been the copper cable that leads to our offices and houses. It's such a shame that when Alexander Graham Bell did all his work on the telephone, he didn't come up with the fibre optic cable as well.

DSL really seems to be the panacea we've been hunting for - it promises to deliver data over the copper phone lines at previously unimagined megabits per second. DSL came into awareness a couple of years back when it was going to be the delivery mechanism for video-on-demand systems.

That fell by the wayside, but now it's back and associating itself with the internet - confirming the old adage, where there is a bandwagon there is a way.

The new thinking from some of the players in the DSL market is that technical limitations mean video-on-demand is unlikely to work. But DSL would be terrific to access the internet. This is all the information companies needed to wage a full marketing assault on the industry.

DSL basically works by using more of a copper wire's capacity than a basic analogue modem. Traditionally, the cable that sits between the phone and the exchange uses only a 4KHz bandwidth. DSL pushes this up to 1MHz.

To date, there are many different flavours of DSL services, including asymmetric DSL, rate-adaptive DSL, symmetric DSL, high bit-rate DSL, very high DSL, and video DSL. Each has its place, but the main technology is asymmetric digital subscriber line or (ADSL).

ADSL moves downstream data from the internet - which to the user is faster than upstream data - away from the user. Hence, the term asymmetric. It achieves download speeds of up to 9Mbps and upload speeds of up to 1.5Mbps over regular analogue phone lines. Because it is using unused frequencies on existing phone lines, you can use the same line for telephone calls.

One of the most attractive things about ADSL is that it bypasses most of the public switched telephone network (PSTN), which often delays analogue modem and ISDN connections with engaged signals, no answer calls, and other general reliability problems.

The one proviso with this technology is that the cable carrying the signal can only support a 1MHz bandwidth for wire lengths of between 2.2 and 3.4 miles. By spreading a signal over more frequencies, a wire can support higher transmission data rates.

In the US, however, this restriction excludes about 40 per cent of all homes since they are based in rural or suburban areas. Therefore, it isn't too surprising that Dataquest predicts ADSL and arch-rival cable modem will be neck-and-neck by the year 2001, with about 1.4 million users each.

And how the US handles this technology is likely to define what happens around the rest of the planet.

It also means that whichever high-speed, high-bandwidth service is to be supplied, it has to effectively reside in the exchange - again, a perfect reason to have DSL lines point to the internet. This is doubly good if services that previously couldn't work via the phone system could be accessed via the internet. However, with higher cable frequencies, the signal is more susceptible to noise and interference, and there are still more than a few interference crosstalk problems to iron out.

Exploiting higher frequencies requires special equipment in the telephone exchange. A DSL system does not use the standard analogue-to-digital converter found on normal telephone lines and, for example, modems. Instead, the carriers use special line cards with digital signal processors (DSPs) that use sophisticated techniques to squeeze out the extra line capacity.

Besides tapping DSP processing power, a more intelligent use of available upper frequencies is required to get data over a copper line at the higher rates. For a start, the voice line element of the signal has to be separated from the data element, and so the system needs a splitter. This allows the phone to work normally and also shunt off the data to the waiting arms of the nearest computer.

The splitter is the cause of yet more bother, namely, how to divide the spectrum and optimise its use. But this is the computer industry, and so, not for the first time, it seems to be going in different directions at the same time.

For ADSL, there are two main directions: carrierless amplitude and phase modulation (CAP), and discrete multitone (DMT) line encoding.

Each is a different modulation system - a method for putting data onto a carrier signal and reading the data at the receiving end. CAP uses a single carrier; DMT uses multiple carriers.

At the moment, about 90 per cent of ADSL services use CAP, but, of course, this doesn't mean it will win out since DMT is the basis of current ANSI and European standards for ADSL.

But that's not the end of it. CAP is, according to some, the best route forward. For a while now, Rockwell has been developing solutions along the CAP route, gleefully rubbing its hands at the prospect of getting to a solution before everyone else. But just like any cheap horror movie (the quintessential model for the computer industry), it's just never that simple.

So it would have been no surprise to anyone that Compaq, Intel and Microsoft - the industry's holy trinity - have got together and, for reasons known only to their marketing arms, plumped for the DMT system.

Earlier this year, the trinity, along with some telcos (including Bell Atlantic in the US and BT in the UK) created the universal ADSL working group (UAWG), which divulged its vision of the future of ADSL, and with the usual industry panache, called it the g.lite specification.

If you listen carefully, you can hear the low moaning from various laboratories around the world as technicians throw their work away.

Rockwell, still bloody after the 56K modem wars, knows what is good for it. It is currently to be seen running after the UAWG at full tilt, desperate to make sure it is with the winning team this time.

Along with the splitter's technical difficulties, there are usability problems that demand it works quickly and efficiently in the home. A lot of the development is to give off-the-shelf functionality - meaning the consumer can buy an ADSL adaptor, simply plug it in and use it.

However, there are even more DSL variations in the works. One is high data rate DSL, or HDSL. Unlike ADSL, HDSL is symmetric, moving data at speeds of up to 2Mbps in both directions.

Other versions on the horizon include ISDN DSL (IDSL), which provides 128Kbps download speeds; single-line DSL (SDSL), with download speeds of up to 6Mbps, but a line distance of 10,000ft; and very high data rate DSL (VDSL), which is restricted to users located within a mile of the phone company's central office, but promises download speeds of up to 52.8Mbps.

DSL is a technology looking for a home and seems to have found the perfect ally - the internet. But the issues surrounding some of the technical problems will have to be addressed sooner rather than later.

Unfortunately, it would appear that the marketing engines of the big corporations involved have slipped into overdrive, and it is becoming increasingly difficult to separate the marketing wishful thinking and real comment.

Microsoft

One of the big players in the ADSL group is Microsoft. It has no product - hardware nor software - though so why is it there? John Billings, senior sales engineer of the internet customer unit at Microsoft, explains: 'Microsoft fits into this because we think it will sell more PCs and NT server software.

If there is better connectivity to the home and the small business workplace, we aren't going to make any money out of DSL hardware. But we do believe it will ultimately increase the adoption of PCs and hopefully CE-type devices, and, in the shorter term, push small business services.' This leads to the obvious question of whether Microsoft's role is as a developer or as an enabler for the market. Billings makes himself clear: 'The latter - DSL has been kicking around for 10 years or so, and the idea is to take it by the scruff of the neck and standardise and package it in a consumer way.

'The initiative that Microsoft has in this is the universal ADSL working group. This includes BT and the idea is to drive the standard. Microsoft was one of the initiators saying "we are all doing work in this area, we all think it can do great things for the future can we get together and make things that work and can be adopted"?'

But what exactly are we going to see from Microsoft? 'We have done some things,' says Billings. 'Most of it is a hardware thing, but there are some soft components and we have created setup wizards, which has traditionally been an awkward area for DSL and requires expert help.

'We will probably be supplying software that will enable you to connect to your ADSL service provider. It will ask some relatively simple questions and at the end of it, your device is properly configured. It may even do the last step of phoning your local service provider - BT in this country - and placing the requisition for the DSL equipment for the exchange.'

So a reasonably unusual role for Microsoft, playing both referee and enabler among the hardware vendors. How long will that last?

Rockwell

If anyone has had to take these changes on the chin, it is Rockwell.

Having started out in one direction, it had to swiftly readjust as the market changed. Nick Burd, technology planning manager of DSL products at Rockwell, talked through it: 'To have ADSL at home you need to have a splitter, which is essentially a glorified filter, separating high-speed data across two wires - one wire going to your PC and the other to your telephone.

'Essentially, it is a pain in the backside because in order to deploy a service, some person has to get into a truck, go to your house, stick this thing on the wall and set it up. If you compare that to the modem model - where you go out and buy a modem, plug it in and off you go - the view to the consumer is that this DSL stuff is rocket science.

It isn't consumer friendly.

'Now, about a year ago, Rockwell took a look at this and decided DSL was going to be the Next Big Thing, but the biggest problem was the splitter, which was going to need a specialised group of people to set up.'

Rockwell came up with a solution that didn't require the installation of a splitter and called it CDSL, or Consumer DSL. 'We announced this technology back in October 1997,' says Burd. 'And about that time, Microsoft, Intel and Compaq were having similar thoughts as us.' But it turned out that although these thoughts were the same, implementation wasn't - the UAWG decided on the discrete multitone (DMT) route, which Rockwell had discounted earlier.

So is this just like the split in the 56K modem market? Burd doesn't think so. 'Well, it's not because we are not fighting it,' he says with some irony. 'There is no way at all that Rockwell will ever hope to come up against the likes of Microsoft, Intel and Compaq and say DMT isn't a good idea - let's be realistic about these things. We will do a DMT-based system and we are putting plans to do that right now.

'We are participating in the UAWG, and since we have been working on splitterless technology for the past 18 months, that gives us a lot of expertise on the pros and cons of how you do it. We hope that we can contribute by making what is not essentially a good technology into something that can deliver the goods.'

All in all, Rockwell knows which side its bread is buttered and isn't going to rock the boat - which seems pretty sensible considering who owns the boat.