Opportunistic Sensing in Task-Oriented Wireless Sensor Network Based on Graph Compressed Sensing

As artificial intelligence and modern signal processing technologies progress, sensor networks often necessitate not collecting information from all nodes in order to effectively perceive and monitor target areas in practical applications. Such progress sets the foundation for Opportunistic Sensing (OS) which is a method engineered to automatically discover and select sensor nodes for efficient data gathering. In this paper, we propose a novel OS algorithm for optimizing node deployment in task-oriented wireless sensor networks. It can efficiently fuse sensed information by partitioning the network into multiple subnetworks and integrating graph compressed sensing with Restricted Boltzmann Machine techniques. Moreover, we employ the Kullback-Leibler divergence to quantify information distortion induced by OS. We also introduce the brainstorm optimization algorithm to improve sensor selection strategy. Experiments demonstrate that the proposed algorithm can efficiently diminish reconstruction errors and enhance network performance compared with classical and recent baseline methods.