Optimal and Efficient Algorithms for Projection-Based Compressive Data Gathering

We investigate the problem of compressive data aggregation in wireless sensor networks. We propose a data gathering scheme using Compressive Sensing (CS) by building up data aggregation trees from sensor nodes to the sink. Our problem aims at minimizing the number of links in the trees to minimize the number of overall transmissions. We formulate the problem of constructing aggregation trees for forwarding the compressed data to the sink as a mixed integer linear program (MILP) and present efficient algorithms to solve the problem. We show that our algorithms have outstanding performance and order of magnitude faster than the optimal model.