Spatial diversity in wireless sensor networks

This thesis focuses on the applications of spatial diversity for (i) distributed data fusion and (ii) improving estimation accuracy in sensor topologies that can be modeled as correlated Gaussian random fields. Distributed sensor networks, are essentially multi-terminal systems. This motivates the application and extension of recent advances in Multiple Input Multiple Output (MIMO) systems to Wireless Sensor Networks (WSNs). The capacity increase that diversity promises translates to energy efficiency and latency reduction in distributed sensor networks. We apply Space Time Block Codes (STBC) for joint transmission and data fusion, a method that enables high rate information retrieval. The fusion schemes are based on orthogonal space-time block codes, i.e. the Alamouti (Quaternion) and the Octonion codes. The reduction in latency of the proposed space-time fusion is demonstrated through comparison with optimal fusion schemes under identical constraints.; In the latter part we consider distributed estimation for correlated Gaussian random fields in wireless fading channels. The optimum estimator is derived based on a Multiple Input Multiple Output (MIMO) abstraction that integrates data correlation with spatial diversity. The estimator performance is characterized in terms of multiuser diversity, channel signal-to-noise ratio and receive diversity of the system. By integrating spatial correlation with receive diversity, we show that the field estimation accuracy can be significantly improved while ensuring energy and bandwidth efficiency.