Distributed anomaly event detection in wireless networks using compressed sensing

Compressed sensing (CS) is an emerging theory that has earned increasing interests in the area of wireless communication and signal processing. It states that the salient information of a signal can be recovered from a relatively small number of linear projections. In wireless networks, anomaly detection is an attractive application. Current research shows that it is promising to apply CS into sparse anomaly network detection, as the number of abnormal events seems much smaller than the total number of nodes. In this article, we firstly propose an ameliorated reconstruction method for abnormal event detection in noise-involved wireless networks, where no prior information is needed. Second, we improve this method to solve a distributed anomaly detection problem considering energy consumption and detection accuracy. Finally, we analyze the performance of our scheme in different conditions. Simulation shows that our detection algorithm proves to be valid and much energy can be saved by the distributed scheme with acceptable performance.