Spatio-Temporal Compressive Sensing-Based Data Gathering in Wireless Sensor Networks

Abstract

Sensory data in many wireless sensor networks feature spatio-temporal correlations, and compressive sensing (CS) plays an important role in energy-efficient data gathering. In this letter, we design a new CS-based data gathering algorithm, utilizing random sampling and random walks to select sensory data in temporal and spatial domains, respectively. Each measurement is obtained by summing the selected data. A novel sensing matrix is also designed based on the adjacency matrix of an unbalanced expander graph. Simulation shows that our proposed algorithm reduces energy consumption by up to 50.0% compared to the existing algorithms in a daily sea surface temperature measurement scenario.