Compressive Sensing-Based Data Aggregation Approaches for Dynamic WSNs
RIS ID
136558
Abstract
Among various data aggregation approaches proposed for wireless sensor networks (WSNs), the one based on compressive sensing (CS) has the merit of low traffic cost. The key step to design a CS-based data aggregation protocol is to construct a measurement matrix {\Phi } based on the network structure and to assign each node a unique column vector of {\Phi }. Assuming an expanded scenario, where some new nodes join in the network, the data aggregation scheme has to entirely re-generate a new matrix {\Phi ^{\prime }} with a larger size to meet the node number and CS property simultaneously. Apparently, it is energy-consuming to reallocate the weight vectors from the new measurement matrix to all the nodes. Thus, we propose an approach which aims to keep the weight vectors of existing sensors unchanged but assign only optimized measurement vectors to the newly added nodes. In order to solve the relevant non-convex optimization problem, two efficient methods with good data aggregation performance are proposed. Numeric experiments validate the effectiveness of our proposed methods.
Publication Details
X. Wang, Q. Zhou, Y. Gu & J. Tong, "Compressive Sensing-Based Data Aggregation Approaches for Dynamic WSNs," IEEE Communications Letters, vol. 23, (6) pp. 1073-1076, 2019.