A Perception Method of Direct Sequence Spreading Spectrum Signal Based on an Improved Residual Neural Network

The problem of detecting spreading-spectrum signals under non-cooperative conditions has always been a hot research topic in the field of communication countermeasures. Traditional detection methods have poor performance at low signal-to-noise ratios. To effectively improve the performance of spread-spectrum signal detection, a spread-spectrum signal detection method based on an improved residual neural network is proposed. First, the correlation peak characteristics of the spread-spectrum signal are quickly extracted by the generalized cross-correlation algorithm. Then, the residual neural network is used as the basis, and the attention mechanism is integrated to establish the network model and analyze and identify abstract features. Finally, the proposed method is experimentally verified on a simulated data set. The results show that compared with the time-domain autocorrelation method, the convolutional neural network method, etc., the proposed method has better detection effect and can effectively detect spread-spectrum signals at a signal-to-noise ratio of -16 dB.