Member Failure Recovery Mechanism Using Context Ontology for Community Computing

Community computing is a new paradigm for building ubiquitous systems which can solve problems by collaborating between entities in ubiquitous environment. Community is a kind of collaborative group and members in the community interact with each other for achieving community goals - that is, seamless interaction is the key issue of the community computing. In this perspective, member failure is one of the most critical issues, because member failure interrupts interactions between members. The purpose of this paper is to solve the member failure problem. We suggest a new replication mechanism that community maintains all member statuses as replicas and replicates member using replica of failed member when member failure detected. And then, replicated member is transformed himself to take part in the community. For the fast and efficient recovery of the member failure, we define some ontologies representing context for community entities as well as replica of the member written in Web ontology language (OWL). As the experimental test, we simulate a plausible scenario on JADE platform and evaluate our approach in terms of cost and availability.