Verifiable middleware for secure agent interoperability

There is an increasing need, within organizations such as the Department of Defense and NASA, for building distributed applications that are rapidly re-configurable and survivable in the face of attacks and changing mission needs. Existing methods and tools axe inadequate to deal with the multitude of challenges posed by application development for systems that may be distributed over multiple physical nodes separated by vast geographical distances. The problem is exacerbated in a hostile and unforgiving environment such, as space where, in addition, systems axe vulnerable to failures. It is widely believed that intelligent software agents are central to the development of agile, efficient, and robust distributed applications. This paper presents details of agent-based middleware that could be the basis for developing such applications. We pay particular attention to the correctness, survivability, and efficiency of the underlying middleware architecture, and develop a middleware. definition language that permits applications to use this infrastructure in a scalable and seamless manner.