Compressive sensing based sparse event detection in wireless sensor networks

We investigate the compressive sensing theory(CS) for sparse events detection and reconstruction in energy-constrained large-scale wireless sensor networks(WSNs). In order to save more energy and prolong the lifetime of the network, we partition the nodes into C sets nearly uniformly in a purely distributed way by using game theory. In each specified time slot, we only wake up parts of the C sets nodes, and set the rest nodes to sleep for saving energy. Based on the proposed sleeping strategy and capitalizing on the spatial sparsity of the event in the local area, we apply compressive sensing theory to gather and reconstruct the sparse signals. The proposed algorithm for sparse events detection is able to efficiently reduce the number of sensors without introducing intensive computation and lose of detection resolution. Especially, the proposed algorithm is mainly based on the greedy algorithms, such as Orthogonal Matching Pursuit, Regularized Orthogonal Matching Pursuit and Subspace Pursuit algorithm. What is more important, based on the game theoretical sleeping strategy, compressive sensing algorithm shows a better detection resolution than the random sleeping strategy. Finally, extensive simulations confirm the performance and robustness of the proposed algorithm under noised environment.