摘要

In traditional compressive data gathering (CDG), many rounds of data gathering are required to make sure the accuracy of reconstruction signal such that most of nodes are required to participate in data gathering. By using the spatial and temporal correlation character of sensor readings, we propose a CDG method based on representative node selection (RNS-CDG) to collect measurement vector via just one routeing tree. By means of principal component analysis (PCA) and frame potential (FP), the proposed method can select fewer representative nodes from all nodes, by which a data gathering tree will be constructed. And then, via this routeing tree, a measurement vector composed of M projections will be received by Sink for recovering signal. It is shown via simulation results that the proposed method can ensure the accuracy of data reconstruction and reduce the transmissions of network.

全文