A VERSATILE ARCHITECTURE FOR THE DISTRIBUTED SENSOR INTEGRATION PROBLEM

The computational issues related to information integration in multise nsor systems and distributed sensor networks has become an active area of research. From a computational viewpoint, the efficient extraction of information from noisy and faulty signals emanating from many sens ors requires the solution of problems related a) to the architecture a nd fault tolerance of the distributed sensor network, b) to the proper synchronization of sensor signals, and c) to the integration of infor mation to keep the communication and the centralized processing requir ements small. In this paper, we propose a versatile architecture for a distributed sensor network which consists of a multilevel network wit h the nodes.(processing element/sensor pairs) at each level interconne cted as a deBruijn network. We show that this multilevel network has r easonable fault tolerance, admits simple and decentralized routing, an d offers easy extensibility. We model information from sensors as real valued intervals and derive an interesting property related to inform ation integration in the presence of faults. Using this property, the search for a fault is narrowed down to two potentially faulty sensors or communication links. In a distributed environment, information has to be integrated from ''temporally close'' signals in the presence of imperfect clocks in a distributed environment. We apply the results of past research in this area to state various relationships between the clocks of the processing elements in the network for proper informati on integration.