Natural Inspiration for Self-Adaptive Systems

  • Authors:
  • Richard John Anthony

  • Affiliations:
  • University of Greenwich, United Kingdom

  • Venue:
  • DEXA '04 Proceedings of the Database and Expert Systems Applications, 15th International Workshop
  • Year:
  • 2004

Quantified Score

Hi-index 0.00

Visualization

Abstract

The emergent behaviour of autonomic systems, together with the scale of their deployment, impedes prediction of the full range of configuration and failure scenarios; thus it is not possible to devise management and recovery strategies to cover all possible outcomes. One solution to this problem is to embed self-managing and self-healing abilities into such applications. Traditional design approaches favour determinism, even when unnecessary. This can lead to conflicts between the non-functional requirements. Natural systems such as ant colonies have evolved cooperative, finely tuned emergent behaviours which allow the colonies to function at very large scale and to be very robust, although non-deterministic. Simple pheromone-exchange communication systems are highly efficient and are a major contribution to their success. This paper proposes that we look to natural systems for inspiration when designing architecture and communications strategies, and presents an election algorithm which encapsulates non-deterministic behaviour to achieve high scalability, robustness and stability.