Self-Organisation

Self-organisation characterises systems which spontaneously order or structure themselves without the need for outside intervention or control. Self-organisation occurs in many systems in physics, chemistry and biology, examples include embryogenesis, the BZ reaction, convection cells and ant colonies.

Understanding the dynamics of these natural systems enables the design of artificial systems of multiple interacting units which will spontaneously form into application-specific structures to given an arbitrary starting configuration.

Examples of applications include systems for frequency allocation in mobile networks, self-organising community formation based on user behaviour and building routing tables for query resolution in decentralised networks.

Algorithms adopting this approach benefit from robustness, scalability, decentralisation and adaptivity. In some cases, problems which are analytically untractable can be solved near-optimally through self-organising processes. The technique is particularly appropriate for systems which must adjust rapidly to changing conditions or handle noisy data sets, where the modeling and explicit organisation of the entire system is impractical, or where identifying relevant structure is beyond the capability of human users.

ArchiveResearch: SWAN, DIET, Flyphones