The future of…

Click here for contact details, other articles and personal details

The future of networks

September 2001

As I write this, BT has been in discussions about selling off our networks, both fixed and mobile. This would have been inconceivable for BT's business just a few years ago to anyone outside the company, but in actual fact the idea goes back to the early 1990s. There are many valuable services that can be run by a telecommunications company but the amount that they use the network has little to do with their value. The most intensive bandwidth hungry services have much less value per bit than most low data rate services. While networks will always be needed, and some profit can be made from running them (otherwise they would be simply switched off), the bulk of the value in modern services lies in the value added to the transmission. Since BT's skills will increasingly lie in squeezing extra value from communications channels of all kinds, provision of the network could easily be outsourced, allowing us to refocus. We could simply buy transmission capacity on a shared network and extract value from the service as a whole, transmission being just one component of service value.

The nature of networking has changed considerably over the years. We used to have a dedicated circuit all the way from one phone to another. A few decades ago it became possible to allow calls to use the same circuit, but the local network still had a dedicated line all the way to the exchange from the phone. Now, many devices can share a link at any stage of the network. The use of packetisation, primarily of course IP, allows any device to take a share of available capacity and to communicate at the same time as many others.

There are a number of threats to the existing business of simply selling capacity across a network. The rapid reduction of cost of transmission makes it very hard to recover investment, since this must be done in a very short time after deployment. If the network is large, deployment takes a long time and becomes very risky. Technologies in development at the moment around the world threaten to bypass paid-for networks altogether for many users for much of the time. Parasitic networks, also known as ad-hoc networks and symbiotic networks, coupled with another new area called software radio, will mean that a computing or communications device will be able to look around for a free network that it can use and access it. If none exists locally, it can initiate the creation of a new ad-hoc network, using any other devices in the neighbourhood. Data can flow hop by hop through to the destination. While there are some concerns about overall performance from such networks, they can certainly be made to work in principle and in the laboratory. It remains to be seen how well they will work in the field. Another way of avoiding charges that may be commonplace is to use a wireless LAN to access a corporate network for onward routing. Most companies do not welcome outsiders onto their LANs, but there are many good business reasons why some enterprises are all too happy with the idea. For example, coffee shops may offer free access to keep people in the shop drinking coffee. Supermarkets may give free access to encourage shoppers to use their supermarket rather than a competitor. Still others may give access in return for advertisements on terminals.

As if being able to network without charge isn't a big enough threat to commercial networks, another potentially bigger threat exists. Real time information and communications have little choice but to use some kind of network, but most services aren't real-time. Some of the 3G hype pictured users downloading videos across the network just before they watch them. This may or may not eventually prove technologically feasible, the jury is still out. But why would someone download a 90 minute video across a link at a few hundred kilobits per second when they could download it into a portable device at tens or hundreds of megabits per second before they leave home? The intelligent use of available storage technology will sort the winners from the losers in the future networking game.

Already, TiVo devices are appearing in many homes. These act as video buffers to allow stop-start watching of TV. Users can freeze TV when they leave the room and start it again when they return. The buffer is a hard disk, so can last hours if need be. The same store can act as a video recorder. The manufacturers sell this as a TV device but the store could work equally well for any form of digital data. A bit is a bit. It is conceivable that as hard disk capacity rises dramatically over the next few years, a home may have a single data store, with several inputs and several outputs. It would accept TV, radio, internet, DVDs, CDs, memory sticks and any other common data storage media as inputs, and be able to write the data very quickly to memory sticks, external devices, or writeable hard media. Using such a device would solve many of the problems associated with data being distributed amongst several home computers, as well as opening up the full potential of the computers to make the most of the various domestic data sources.

But one of the other business characteristics of TiVo highlights the potential business in this area and shows that it is not so much a threat as an opportunity. As well as selling the storage device, they also sell a subscription to a directory service. Via a downloaded programme guide, the device can ensure that the correct programme is recorded, even if it is delayed. It also learns what sorts of programme the user watches so that it can ensure that it records the most appropriate programmes even without explicit instructions. It is so useful that Sky have started selling one such device, even though it has the capability to bypass advertising, one of their primary revenue streams.

But valuable though this service is today, it is just a hint of what is to come. The future is polluted by vast quantities of information. The volume of information in the world doubles every year and the rate is accelerating. We will spend heavily on service that filter this vast pool of information and give us what we want, when we want it. Herein lies the key to tomorrow's network revenue.

Instead of a dumb network that just sets up calls on an explicit instruction, between two points, we will have a much more intelligent network that tells us what we want when we want, and puts us proactively in touch with people that it knows we will want to talk to. It will know where we are, and where our friends are. It will know each of our schedules, so can arrange our lives to help us make more of our time. We will earn more and have more free time to spend it, and have more fun doing so. We will be only too happy to pay for the consequential improvements in the quality of our lives.

The storage based network of the future will have many intelligent components. It will still need a fixed and mobile network for data delivery, even if it only uses them occasionally. It will need a wide range of smart terminals, each of which might have a very different range of capabilities. The network will know whether it can send a video signal or just audio or whether it needs to make a voice to text translation. It will know the capabilities of other terminals near to where you are, and will be able to negotiate with those terminals so that it can send appropriate information to you even if you aren't carrying a suitable device on your person.

Data stores will be ubiquitous. It will never be possible to store all of the world's data on a single device. Today there are 60 exabits of information in the world, and a state of the art hard drive can hold about a terabit, but the information is growing faster than the density of storage (this is only possible because the number of storage devices is also increasing). However, the amount of useful data in the world is increasing only slowly. So the most useful component of the future storage net is the filter. It needs to know what we like and want and to be grab this tiny subset out of the media and keep it on our personal store. When a new DVD pops in the post in a few years time with the latest films, it will only extract the ones we are likely to want to watch. It will sit like a barnacle in the information flood across the terrestrial, satellite, and cable nets and extract what we want. Our personalised data will fit in our home data store, and be accessible to us throughout the home. When we want to go out, we will drop our portables onto store and it will upload what we are likely to need for the day ahead. It will download appropriate video and music, maps and tourist data for the places we are going, all the emails and data for the events we are attending, anything it thinks we might need. We will still access the network for some data, but most of it will be in out portables already.

The service that filters and provides us with such a useful travelling companion will be worth far more to us than any amount of dumb communication. We can avoid paying for communications if we try, but we won't be able to cope when out storage based network service is offline.

Author Biography

Ian Pearson graduated in 1981 in Applied Mathematics and Theoretical Physics from Queens University, Belfast. He spent four years in Shorts Missile Systems, in many different disciplines from mechanical engineering to battelefield strategy simulation. He joined BT Laboratories in 1985, analysing the performance of computer networks and protocols. After two years he moved to the Local Access Division where he helped develop ATM transmission over optical networks and invented Addressed Time Slicing. In 1990 he moved to the Network Studies Unit, where he worked on evolution of broadband networks and services. Since 1992, he has worked in the Advanced Research Department, in the Cybernetics, Networks and Mobile Systems groups. He now concentrates on mapping the progress of new developments throughout information technology, considering both technological and social implications. He currently works as BTexact's futurologist. Ian was awarded the Best Paper award at the 1993 FITCE Conference for his first externally published paper and has since received seven other awards for published papers, including the IEEE Benefactors premium in 1994 and the IBTE Journal Best Paper Award in 1996. He is married with a seven year old girl called Rachel and enjoys music, swimming and reading in those few moments left after playing with Rachel.