Scientists at British Telecommunications Plc’s Martlesham Heath laboratories near Ipswich are working on a project to help route traffic around telecommunications networks based on the study of ants. The scientists don’t actually bring in ant farms and watch them, but there is a vast amount of scientific research available on how simple creatures such as ants communicate and co-ordinate to the point that they appear quite intelligent when all working together in the nest. Based on the ant studies they are coming up with a new and simpler approach to writing programs in which they break up a large monolithic program into smaller and simpler reactive blocks, resulting in what they call ‘mobile agent’ software. Using conventional object-oriented programming, the resulting code is much shorter and easier to write and understand. The small programs allow for a certain amount of random behaviour as it is not all implicitly pre-programmed. Instead, if a program is set a task, it looks for its own way of solving the problem in hand. In effect ‘parent mobile agents’ are sent out to scout for trouble-spots on the network and these spawn ‘worker agents’ at those points. The new approach crosses the boundaries of salient computing issues, for example, that of central versus distributed control. The mobile agents can blur that boundary and emulate central control and also emulate distributed control along any point of that spectrum. Project researcher Simon Steward believes that such software will come into practical use surprisingly quickly – in as little as two to three years. There is potential for huge cost savings to be made in terms of both the numbers of lines of code needing to be written being reduced and a reduction in the potential risk of a total network failure as the new systems more effectively route calls.
