A new adaptive compressed sensing algorithm for Wireless Sensor Networks

In this paper, a new adaptive compressed sensing algorithm for Wireless Sensor Network (WSN) was proposed. Power efficiency is an important requirement in WSN, however, measurement matrix used in classical compressed sensing is always dense, which can not satisfy this constraint. In the proposed algorithm, a new metric named total coefficients power is defined to guide the node selection to build a sparse additional projection vector, and the differential entropy is adopted to determine the coefficients. Simulations show that this new algorithm can obtain good reconstruction performance while reducing the communication cost.