Matrix completion based sensor selection strategies in wireless sensor networks

Wireless sensor networks (WSN) have been widely used in many fields. Currently, compressive sensing (CS) is mainly adopted to compress, transmit and recover sensing data in large scale WSNs in order to decrease the amount of data to transmit and reduce the energy consumption of data transmission. In this paper, we deal with sensing data recovery via matrix completion (MC) which is a hotspot nowadays. Compared with CS, we decrease the amount of data to collect directly rather than collecting more data and then compress them for transmission. To further decrease the amount of data to collect and transmit, we proposed an information-based sensor selection strategy in WSNs to idle the redundant sensors and let them work in sleeping mode. We simulate with practical dataset based on Singular Value Thresholding (SVT) algorithm and the results show that we can reduce the number of working sensor nodes from 70 to 41 within a certain error tolerance. The energy consumption of the network reduced significantly up to 41.4%.