An energy efficient sampling method through joint linear regression and compressive sensing

The wireless sensor networks (WSNs) are composed by the energy limited nodes. Data transmission can be decreased by exploiting both intra- and inter-signal correlation for saving energy consumption of nodes. In this paper, a dynamic clustering algorithm was proposed by exploiting inter-signal correlation. The nodes can be divided into clusters according to their data's linear correlation by our proposed clustering algorithm. Based on clustering, we also proposed an energy efficient sampling method through joint linear regression and compressive sensing (CS). The proposed method can accurately reconstruct data of nodes by using significantly low sampling rate at each sensor. Our results based on real data sets indicate a reduction in sensor sampling by up to 71%.