The distributed system based on agent entities is a system that is subjected to failures because of the complexity that characterizes the structure of each entity and to the various interactions between the different agents. Although, there are some approaches that deal with fault failure, there is still a need of models that objectively address the area of fault tolerance, especially when the existing approaches cover fault tolerance using traditional methods; these methods are based on the underlying operating system and communication. In this paper, we propose a model based on the sentinel approach to deal with conflict between agents, and integrates fault tolerance directly into the language representing the behavior of the agent. The model proposes three operating modes to better specify the behavior of the agents: normal mode, temporal tolerance mode and degraded mode.
展开▼
