Distributed Source Compression And Cooperative Communication In Wireless Sensor Network

The main purpose of this paper is to design a universal practicalparadigm for distributed source compression, which is motivated by theclassical information theory and Slepian-Wolf Coding. We will not requirethat the source must be stationary and ergodic. The existence of sourcesequences’ entropy, which can be calculated by statistical method, makes itpossible to achieve independent source compression at the cost of quite asmall feedback expense. However, the measurements for optimal transmissionof compressed source differ from those of multi-access channels, as theultimate goal of the former is to enable the data-collecting node to correctlyrecover the original data meanwhile with minimum distortion.We are aiming at compressing data and transmitting information insingle-hop networks. First of all, a system model of two source nodes and onedestination node is established. Our design for distributed source compressioncan be interpreted as the scheme of independent encoding and joint decoding,from data training, to reducing redundancy, to feedback the needed encodingrate back towards compressing node.Further, under the system model mentioned above, we will conductresearches to fight with challenges in data processing, distributed sourceencoding, feedback link analysis, and joint decoding step by step. The natureof data’s correlation in wireless networks will be explored to help reduceintrinsic information redundancy. On the premises of data correlation wepropose methods to determine the optimal parameters for data processing ofboth memoryless and memory models. Beyond that, bounds for parameters aregiven.In the following, we focus on the three challenges involved in theoptimal cooperative link design for the three-node distributed sourcecompression system. The challenges are mainly, but not limited to, reliable end-to-end communication, appropriate measurements and switches betweencooperative source nodes. We choose minimum rate distortion as primemeasurement, based on which we propose optimal power allocation and threeschemes for node switching. In the final analysis, we shed light on the likelydeclining trend of battery power as discharge proceeds on. Contrast ofsimulation is given to show the difference between constant and decliningpower provision.The main innovation of this thesis lies in the interactive strategy ofindependent encoding for distributed sources, where three nodes wereincluded in the system model and minimum rate distortion was adopted asmeasurement, in addition, the bounds for the strategy’s parameters werederived. Contribution also rests with three proposed strategies for noddingswitching with the consideration of the influence that sources’ transmittingpower decline imposed on cooperation.