Improved cyclostationary feature detection based on correlation between the signal and noise

Abstract

With the presence of noise floor, characteristic points of signals are submerged easily in conventional cyclostationary feature detection. In this paper, a new cyclostationary feature detection method based on the correlation between signal and noise is proposed for detection of OFDM based primary users. The proposed scheme aims to improve the noise robustness through making the utmost of the information in the cyclic spectrum of the received signals. Meanwhile, in order to obtain the clearer feature, the signal processing technology is used to deal with the cyclic spectrum. Finally, the classification detection is realized by machine learning. Extensive simulation results demonstrate that proposed method provides superiority detection performance compared to the conventional cyclostationary feature detection, especially in low SNR environment.