Abstract

A new method to construct an adaptive sparse basement based on compressed sensing is proposed. By the structure characteristic of time-frequency parameters, Gabor over-complete dictionary is divided into several atomic sub-libraries represented by significant atom. MMP algorithm is used to select the optimal sub-library and from which to choose the optimal atom. Replacing inner product operation with cyclic correlation operation by FHT increases the speed of basement construction. The simulation results show that the basement can make the sensing signal sparse effectively and recover the raw data in high precision. The method is both reliable and feasible, which has very important theoretical significance to the further development and practical application of WSN.

Paper Details