Impact of correlation in node locations on the performance of distributed compression

Abstract—In this paper we analyze the impact of correlation in both node locations and sensed phenomenon on the perfor-mance of distributed compression. In particular, we consider the optimization of Wireless Sensor Networks (WSNs) in realistic environments. Distributed Source Coding (DSC) is a compression technique performed by, for example, data gathering applications in which energy consumption and bandwidth usage are important factors. In this work, we consider different node deployment strategies and include in our study models of sensed data with different correlation types in order to model more realistic WSNs. Our evaluation takes into account the transmission overhead and signal processing costs associated with distributed compression. The energy model applied makes use of measurements obtained from real experiments. From the analysis we derive a novel metric which is essentially based on correlations of node locations and sensed phenomena. This metric enables the performance estimation of WSNs apply-ing DSC prior to the costly real deployment. Furthermore, we show that significant average energy savings and average lifetime extensions are possible using DSC under dif-ferent node deployment strategies and different data correlation structures. I