On fundamental limits of scalable sensor networks

The goal of this thesis is to lay out the information theoretic limits and develop the cross layer understanding, from the physical layer to network layer, involved in the design of the data gathering systems such as sensor networks. Such decentralized information processing systems are finding applications ranging from habitat monitoring to entertainment.; This thesis begins with the scalability question in wireless ad hoc networks, i.e. behavior as network size grows. In particular, the otherwise inherently non-scalable ad hoc networks can be made scalable either by encouraging local communications or by providing extra resources such as bandwidth to each node in the network. Both static and controllably mobile networks are considered. Next, the design criteria, algorithmic techniques and information theoretic analysis for scalable sensor networks are developed.; The implications for sensor network architecture are then analyzed by introducing the concept of spatial fidelity as the design criterion. The algorithms for the decentralized data fusion to facilitate the acquisition, processing and dissemination of information in sensor networks are presented. Design questions such as optimal sensor density and cooperation sensors strategies are also addressed while analyzing the distortion/density tradeoff. The suboptimal but scalable interpolation strategy is stated in order to explore network density tradeoffs in the presence of measurement error.; Fundamental performance limits are then discussed for the data gathering systems for Gaussian sources and channels. The rate distortion bounds are derived for such systems both with and without considering the sensing channel. The bounds for the m-helper, Berger-Tung and CEO coding systems are derived. An upper bound on a cooperative coding system comprised of two transmitters and two receivers is then derived. Here, the different permutations of cooperation among the pairs of transmitters and receivers are exploited, where the data from each transmitter is meant for both the receivers.; Finally, moving above physical layer, a higher layer abstraction with practical implementations in the above described networks is presented. Optimization problems and heuristic algorithms that guarantee the quality of service (QoS) in wireless sensor and ad hoc networks are formulated for this purpose.