Distributed data fusion and information processing in wireless sensor networks

One of the challenges in designing wireless sensor networks (WSN) is their stringent energy constraints. In a WSN, a major part of the energy is consumed by inter-sensor information exchange due to the harsh wireless communication environment. In this study, we address the data fusion and information processing in WSNs, in which energy efficiency the major performance criterion. The problem is approached and solved from both theoretical and practical perspectives. On a theoretical level, we investigate the network source-channel communication problem to reveal the fundamental tradeoff between energy consumption and source reconstruction performance. On the practical side, we design practical distributed signal processing algorithms to approach the performance bounds revealed by our theoretical findings.; In the first part of the thesis we consider the distributed signal processing in WSNs by devising various quantization-estimation and quantization-detection algorithms. Special focus is placed on universal schemes in which the knowledge of data distributions is not required. We then incorporate noisy communication channels in the problem formulation and study the problem of distributed estimation under energy constraints. Both digital and analog approaches are considered. The issues studied in this part include the optimal power allocation, estimation diversity, comparison of analog and digital approaches, and the impact of multi-access schemes.; In the second part of the thesis we study the source acquisition, data communication, and final fusion in a WSN from an information-theoretic point of view. We first give the optimal rate allocation for the vector Gaussian multiterminal source coding, and then provide an improved lower bound for its sum-rate distortion function. Secondly, we extend the Shannon's source-channel separation theorem in some network cases and establish that for the multiple access channel with orthogonal multi-access, the optimal cost-distortion tradeoff can be achieved by separate source and channel coding. The optimal coding schemes in Gaussian sensor networks with orthogonal or coherent multi-access are also discussed.