Data Aggregation Using Distributed Compressive Sensing in WSNs

Abstract

Abstract The theory of Distributed Compressive Sensing (DCS) is best suited for Wireless Sensor Network (WSN) applications, as sensors are randomly distributed in the area of interest. The inter-signal and intra-signal correlations are explored in DCS through the concept of joint sparse models. In this paper, we have analysed joint sparse models and reconstruction of the signal has been done using joint recovery techniques. The reconstruction performance is achieved using joint recovery (S-OMP) and separate recovery (OMP) mechanisms utilising synthetic signals that possess the inherent qualities of natural signals. Further, we employ DCS on real data to evaluate the data reduction. DCS proves to be a better data aggregation technique depending upon the amount of intra and inter correlation between data signals. Simulation results shows that, even in less sparse environments, DCS performs better than separate recovery, which is well suited for real signals. Simulations also prove that with DCS, we can further reduce the number of measurements (compressed vector length), required for data reconstruction as compared to separate recovery. It has been shown that nearly 50% reduction in the data required for reconstruction in case of synthetic signals and 27% in case of real signals. KeywordsDistributed Compressed Sensing (DCS)Temporal-spatial correlationJoint sparsityJoint recoveryJoint Sparse Models (JSM)Simultaneous OMP (S-OMP)Separate recovery