A Novel Strategy for Wideband Spectrum Sensing Based on the Time-frequency Evolutionary Clustering Algorithm

摘要

Aiming at the sub-Nyquist sampled signal with several sub-bands, a Time-frequency Extremum Function (TFEF) is proposed to extract the Enhanced Typical Sparse Time-frequency Feature (ETSTF) of the signal in order to solve the imbalance problem of time-frequency feature extracting for various sub-bands with different SNR(Signal-to-Noise Ratio)s. And an evolutionary clustering algorithm is developed for the ETSTF data set wiping off the false classes' negative effect of idle sub-bands. Based on the clustering results, a novel strategy will be established subsequently for the wide-band spectrum sensing. Simulation experiments indicate the following. 1) The same share have the time-frequency points of various sub-bands with different SNRs extracted by TFEF, which means that TFEF enhances the time-frequency feature of the sub-band with lower SNR. 2) The time-frequency points of busy sub-bands have obvious clustering character. 3) The evolutionary clustering results neglect the idle sub-bands information and have clustering centers in the busy sub-bands.