Events localisation and estimation in wireless sensor networks using compressed sensing

Events localisation and estimation are two major applications of wireless sensor networks (WSNs). However, owing to the noise of wireless sensing and transmission, it is difficult to guarantee the detection accuracy, especially in multiple events scenarios. This paper proposed a multiple events detection method based on compressive sensing (CS) to efficiently achieve the positions and the values of events simultaneously. Moreover, the high redundancy between the sparse signals of adjacent time slots is taken into account by distributed compressive sensing (DCS) to improve the detection accuracy. We also proposed an adaptive CS reconstruction algorithm to deal with the variable sparsity level in practical WSN scenarios. The simulation results show that the proposed method not only outperforms the traditional decentralised detection methods using Bayesian, but also perfectly handles the unknown sparsity level situations.