Role Reconfiguration Based on Decision-Making for Fault-Recovering in Multi- Agent Systems

Distributed hardware systems can be considered as teams of distributed cooperative agents. Thinking this way, will power the designers to develop new agent-based ideas to increase systems’ fault tolerance. In this paper, recovering the system from potential faults by helping the faulty agents in performing their tasks is considered. Besides, each agent, through two phases of distributed decision-making process decides if it can help the faulty agents by undertaking their tasks. If it is possible, the helping actions will be started. The developed ideas are implemented in a simulated Distributed Control System. As it is shown, the proposed distributed fault-clearing method through reconfiguration of the agents’ roles to recover the lost capabilities is very effective. In the presented method, there is no central agent, sentinel or broker to observe the agents and redistribute the tasks among the agents to clear the fault. In fact the system is totally distributed and each agent takes proper actions based on a designed cooperation strategy to clear the fault. The presented methods are tested on a simulated distributed hardware system.