Exception Diagnosis for Composite Service

Since accurately identifying the error source which causes the exception is important, the problem of diagnosis for composite service becomes one of the key issues in the adaptive service composition. This paper proposes an approach for diagnosing composite service. In this approach, the uncertain casual relation between the exception and the service is formed by the way of computing the error propagation degree. The diagnosing algorithm which is based on the error dependent matrix is established. The diagnosis can be achieved by fuzzy reasoning. This approach can preserve the efficiency and accuracy of the diagnosis for composite service, which can support the self-adaptation of the composite service effectively.

EVALUATING PERFORMANCE AND ACCURACY OF A SENTINEL-BASED EXCEPTION DIAGNOSIS APPROACH FOR P2P AGENT SYSTEMS

Using open Multi-agent systems (MAS) to represent a peer-to-peer (P2P) organization is a complex application of distributed artificial intelligence. These systems are designed to create networked applications where each peer node contributes to the overall functionality of the application. Each agent in such systems acts as a peer and has no fixed role. In other words a peer may assume the role of either service provider or service consumer in a given interaction. The dynamics of these networked applications make them vulnerable to different kinds of exceptions. Also the absence of centralized control and changes in organizational structure gives rise to unpredictable exceptions. It becomes essential to have some exception diagnosis mechanisms in place to be able to diagnose the cause of such exceptions while preserving the autonomy of the peer agents. These mechanisms do come with some overheads. In this paper we present an evaluation of the application of our proposed sentinel based approach to exception diagnosis in P2P based MAS and also discuss the trade offs that arise in using a sentinel based approach to exception diagnosis in such systems.