Compressive data gathering using random projection for energy efficient wireless sensor networks

Abstract

This paper proposes a novel data gathering method using Compressive Sensing (CS) and random projection to improve the lifetime of large Wireless Sensor Networks (WSNs). To increase the network lifetime, one needs to decrease the overall network energy consumption and distribute the energy load more evenly throughout the network. By using compressive sensing in data aggregation, referred to as Compressive Data Gathering (CDG), one can dramatically improve the energy efficiency, and this is particularly attributed to the benefits obtained from data compression. Random projection, together with compressive data gathering, helps further in balancing the energy consumption load throughout the network. In this paper, we propose a new compressive data gathering method called Minimum Spanning Tree Projection (MSTP). MSTP creates a number of Minimum-Spanning-Trees (MSTs), each rooted at a randomly selected projection node, which in turn aggregates sensed data from sensors using compressive sensing. We compare through simulations our method with the existing data gathering schemes. We further extend our method and introduce eMSTP, which joins the sink node to each MST and makes the sink node as the root for each tree. Our simulation results show that MSTP and eMSTP outperform the existing data gathering schemes in decreasing the communication cost and distributing the energy consumption loads and hence improving the overall lifetime of the network.