Distributed Processing Algorithms for Wireless Sensor Networks Having Fast Convergence and Robustness Against Coupling Noise

Designing sensor networks with decentralized and autonomous decisions capabilities, i.e., without the need to send all the collected data to a fusion center, is a big challenge that is receiving considerable attention. One of the major drawbacks of distributed algorithms is their iterative nature. This makes them prone to an energy consumption that depends on the convergence time and on the power transmitted by each node to guarantee the network connectivity. Furthermore, in a realistic environment, the interaction among sensor is inevitably corrupted by noise which affects the final decision. In this work, we describe decentralized algorithms for implementing various processing tasks, from spatial smoothing to distributed decision, characterized by fast convergence properties, for a given network topology, and resilience against inter-sensor communication noise.