A nature inspired multi-agent framework for autonomic service management in ubiquitous computing environments

Abstract

This paper describes the design of a scalable biomimetic framework that addresses several key issues of autonomous agents in the management domain of complex ubiquitous service-oriented networks. We propose an autonomous network service management platform - SwarmingNet, which is motivated by observations of the swarm intelligence in biological systems (e.g., termite, ant/bees colonies, or locusts). In this SwarmingNet architecture, the required network service processes are implemented by a group of highly diverse and autonomic objects. These objects are called teleservice solons (TSSs) as elements of teleservice holons (TSHs), analogue to individual insects as particles of the whole colony. A single TSS is only able to pursue simple behaviors and interactions with local neighbors, however, a group of TSSs have the capabilities of fulfilling the complex tasks relating to service discovery and service activation. We simulate a service configuration process for multimedia messaging service, and a performance comparison between the bio-agents and normal agents is analyzed. Finally, we conclude that through bio-swarming intelligence behaviors, this infrastructure develops the enhanced self-x capabilities which give IP networks advantages of instinctive compatibility, efficiency and scalability