A Novel Signal Detection Scheme Based on Wavelet Packet Denoising and Cyclic Spectrum Theory

This paper propose a novel scheme combining wavelet packet denoising and cyclic spectrum feature to detect signal. Since many non-stationary signals existing in communication system have cyclostationnary features, signals can be detected by taking advantage of cyclic spectrum. But when the signal spectrum has more complex structure, the noise component is too large or noise itself has cyclostationary features, the demodulation effect of cyclic spectrum is not satisfactory. In consideration of the limitation of signal identification by employing cyclic spectrum signal identification, this paper introduces a novel scheme combining wavelet packet denoising and cyclic spectrum on the basis of the study of cyclic spectrum principles and the spectrum correlation characteristics of several typical communication signals. In this method, wavelet packet is adopted to delimitate noise using soft threshold and signal is identified according to the feature of its cyclic spectrum afterwards. Simulation results indicate that the method has good noise suppression capability, as well as an excellent signal recognition capability.