Distributed Compressive Data Gathering Framework for Correlated Data in Wireless Sensor Networks

Abstract

In wireless sensor network existing spatial (inter) and temporal (intra) correlation causes redundant data in the network. Exploiting the existing spatial (inter) and temporal (intra) correlation to effectively compress data thereby reducing redundancy to reduce data traffic in network is of prime concern. In this paper, we propose a 2-D distributed compressive data gathering framework to reduce redundancy in sensor network data. We also analyze the performance of the proposed scheme with random DCT and DFT measurement matrices on real data sets of sensor readings with different sparsity. Results indicate that high compression can be achieved with negligible mean square error in recovery from far fewer number of samples than the traditional Nyquist rate at the sink thereby enhancing network life time to large extent in large scale wireless sensor networks. Also, the recovery performance improves depending upon the sparsity measure and the measurement matrices used for compressing the data.