A. Wils, P. Rigole, Y. Berbers, and K. De Vlaminck (Belgium)
ambient intelligence, component based development, em bedded systems, resource management, ...
This paper introduces a contract-driven resource manage ment process and supporting middleware framework. We aim at supporting mobile computing tasks as they appear in the areas of Ambient Intelligence (AmI). As available resources in AmI environments fluctuate frequently, the resource usage of AmI tasks must be extremely flexible. The presented component based methodology permits an optimal flexibility in resource distribution and allocation through the use of resource contracts. User tasks specify their QoS needs in parameterized contract propositions that are used as the basis for a ne gotiation process. Once a contract has been validated and signed by the middleware, it can be monitored. The sup porting decision logic is built around a Java-based expert system. Negotiation and monitoring rules detect misbehav ing tasks and environmental changes. Renegotiations result in task adaptations or more drastic measures, such as partial task relocation.
Important Links:
Go Back
