Plus d'informations ici
Replication heuristics for agents fault tolerance: a plan-based approach
Intervenant(s) : Alessandro DE LUNA ALMEIDA (LIP6)The growing importance of multiagent applications and the need for a higher quality of service in these systems justify the increasing interest in fault-tolerant multiagent systems. In this thesis, we propose an original method for providing dependability in multiagent systems through replication. Our method is different from other works because our research focuses on automatically building an adaptive and predictive replication policy. This replication policy is determined by taking into account the criticality of the agents and the reliability of the machines. We define the criticality as a value (evolving in time) associated to each agent in order to reflect the effects of its failure on the overall system. This value is calculated using the plans of the agents, which contain the collective and individual behaviours of the agents in the application. We also propose the underlying mechanisms of allocation and placement of the available replication resources. For that, we formalize the problem of resource allocation which consists of deciding which replication resources must be allocated to each agent in order to maximize the reliability of the system. Moreover, the set of replication policies applied at a given moment to the agents is fine-tuned gradually by the replication control module so as to reflect the dynamicity of the system.
thomas.genin (at) null