Improved SVM Communication Signal Recognition Based on Information Geometry Denoising

Abstract

Aiming the problem of low accuracy of communication signal recognition by traditional manual feature extraction, an improved SVM recognition method based on information geometry denoising is proposed exploiting the support vector machine (SVM). The proposed method obtains the time-frequency images of different communication signals through the Choi-Williams distribution (CWD) time-frequency transform, and uses the geometric ground distance to accurately measure the difference between pixels for denoising. Then, the AlexNet is used to extract features from the time-frequency maps. Finally, by using the improved SVM based on the information geometry, the classification of communication signal is made to achieve effective classification and recognition. The simulation results show that the recognition rate of the proposed method achieves more than 97% at 0 dB signal-to-noise ratio (SNR). In addition, the method is still effective in the case of small samples.

Keywords

• |information geometry|image denoising|communication signal|modulation identification|support vector machine(svm)|ground distance measurement|alexnet