Autonomic Computing for Pervasive ICT — A Whole-System Perspective
BT Technology Journal
| Hi-index | 0.00 |
The traditional 'Von Neumann' computing architecture is serial and digital. This way of organising things has proved to be very powerful and has allowed spectacular progress in computation, riding on the back of spectacular increases in speed of the central processor.Though the Von Neumann model has outstripped futurologists' dreams in its favoured domains, it has failed to meet expectations in others. Highly parallel, asynchronous, distributed problems highlight its weaknesses. Such complexity is inevitable in today's highly connected networks and dynamic environments with information asymmetries and delays, and myriad interactions among components. There are two ways to address the shortcomings in difficult but important problem domains like these. One is to increase the speed of traditional computing. This is being done, with year-on-year increases in hardware and software performance. The second is to seek new computational architectures which address problems more efficiently. This is also an active research area, and it is the subject of this paper.We are drawing inspirations from nature to deal with the twin challenges of parallel asynchronous problem domains and network system complexity that characterise the information age.