Formal Modeling and Verification of Multi-agent System Architecture

In a multi-agent system, the multiple distributed intelligent agents interact with each other to solve problems. To guide the development of multi-agent system, the multi-agent system architecture would provide a framework. The specific multi- agent system can be customized from the multi-agent system architecture, which does not need to rewrite the construction. In order to satisfy the failure recovery property of multi-agent system, we propose dependable multi-agent system architecture with fault tolerant mechanisms. The PVS formal language is used to build a system architecture, which can provide common patterns and idioms to the system developers. In order to satisfy the reliability requirements, the powerful PVS theorem prover can be used to analyze the high reliability property of the proposed architecture.