Towards Distributed Sensor Networks: An Extended Abstract

Consider a network consisting of a multitude of sensors distributed over some region. When a target crosses the region, observations will be taken by many sensors. These observations at the network nodes are noisy, partial, and have a varying degree of credibility. Therefore although each individual sensor may not have enough information the total system may possess enough information to track the targets. However the total information is distributed in the whole system and cannot be collected or processed by any single node in realtime due to limited communication and processing resources. Nor should it be centrally processed due to reliability and vulnerability considerations. The Distributed Sensor Networks (DSN) problem consists of building a sensor-communication-processing architecture to fuse noisy and partial observations from a multitude of geographically distributed sensors of varying credibilities. The program involves challenging problems of pushing the boundaries of signal processing technology, artificial intelligence and computer communication technology and integrating these disciplines. The objective of this paper is to present some of the interesting open problem areas arising in DSN