Adaptive Agent-Based Service Composition for Wireless Terminals

Software agents are one of the building blocks of ambient intelligence and pervasive computing. Adaptation to changes in the execution context is necessary in order to provide continuous and smooth provision of services for wireless clients. In this paper, we present a system for adapting a service consisting of multiple agents based on application, terminal and communication models and profiles. The system adapts and composes the service by finding the best combination of local and external agents to contact. The context models are based on experimentation with actual devices and environments. We present a theoretical cost model for runtime service composition, and examine experimental results that are based on an example scenario.