Collaborative sparse signal recovery in hierarchical wireless sensor networks

This paper investigates the design choices and implementation schemes for information fusion among cluster heads in a large-scale hierarchical wireless sensor network. Two main issues addressed are: whether to choose centralized processing with aid of a fusion center or decentralized collaboration among cluster heads, and for the latter choice, how to collaborate. Based on a sparse signal recovery problem arising from an environmental monitoring application, we propose a decentralized collaborative decision-making algorithm for cluster heads, and compare it with the centralized scheme. Our observation is: when the number of sensors within each cluster is quite large to induce a large amount of data, and the cluster heads are subject to multi-hop communications due to limited communication range, the collaborative algorithm is superior to the centralized one in terms of communication load and energy efficiency.