Distributed Fusion in Sensor Networks with Information Genealogy

Distributed sensor networks seek to enable adaptive and cognitive behavior in networked information systems. These networks will exhibit truly ad hoc behavior as they adapt in situ to maintain or optimize operations under various conditions. Network topologies and membership may change in response to unpredictable variations in conditions such as spectrum availability, link conditions, power and energy constraints, latency, and routing. As a distributed system of devices, networks must support truly decentralized information exchange, and fusion. Under the ONR Grant: #N000140711211, George Mason University has been developing innovative mathematically rigorous methods for combining data from multiple sources to provide the best estimate of objects and events in the battlespace. Specifically, the key challenge for this research is to develop autonomous fusion algorithms designed for ad hoc wireless network operating under severe communication constraints. These algorithms must be able to scale to large numbers of entities and to combine many disparate types of data. This distributed fusion methodology is both analytically tractable and can be readily implemented in a distributed and autonomous manner. The method is grounded in set-theoretic derivations of information fusion where we develop information genealogy to provide a global view of distributed fusion events for each agent under adverse operating conditions.