Robust reconstruction model for compressive data gathering in wireless sensor networks

In practical applications of compressive data gathering (CDG) in wireless sensor networks (WSNs), the reconstruction accuracy is low when observation and measurement process are corrupted by outlier and noise. To eliminate the influence of outlier and noise, robust reconstruction model (RRM) based on sparse characteristic of outlier is presented in this paper. BPDN algorithm is applied to solve the proposed model, and outlier detection algorithm is designed to find the position of outlying sensor readings. In addition, simulation results show that BPDN outperforms OMP in solving the proposed model. And the reconstruction error of BPDN based on RRM is below 0.1, and the outlier detection accuracy is nearly 90% while the number of measurements is only half of the total number of sensor nodes.