ECO[sbnd]CS: Energy consumption optimized compressive sensing in group sensor networks

Abstract

Compressive sensing (CS) is a widely employed technique in sensor networks for energy-efficient data transmission. In recent years, the group-based network structures, e.g., regionalized and clustered networks, have been proposed to work with compressive sensing to reduce the energy cost of boundary sensors. Studies in previous literatures including exploring the relationships among samples in sensor groups, the techniques for grouping, etc. However, several issues may surface after the group structure is established. To extend the state-of-the-art techniques, we propose an energy consumption optimization approach based on CS, ECOCS, in group sensor networks. Three challenges are addressed in this paper: 1) we show the design principle of group measurement matrix and analyze the expected size of measurements; 2) we present two schemes to obtain candidate sensors that facilitate group collector election and cost reduction when establishing routing schemes based on hyperbolic Ricci flow; 3) we give the reachable probability of accurate reconstruction to avoid unnecessary sampling. The experiments demonstrate that our solutions to these challenges are superior to existing approaches.