Time-Frequency Domain Hybrid Spectrum Sensing Method Based on Deep Learning

Driven by the rapid growth of wireless communications and the increasing scarcity of spectrum resources, advanced spectrum sensing methods have become essential for efficient spectrum utilization. In this paper, a CNN-LSTM integration architecture optimized for dynamic wireless environment is proposed, which extracts frequency domain features through CNN and models the timing dependence of LSTM, so as to improve the detection performance under low signal to noise ratio (SNR); In addition, this study also optimizes the model parameters, including the size of the convolutional kernel, the number of layers, the activation function, etc., to adapt to different signal environments; In terms of time-frequency domain feature fusion, a joint feature extraction mechanism is introduced to enable the model to make full use of the time domain and frequency domain information of the signal. To validate the efficacy of the model, training and testing datasets were synthesized using MATLAB simulations, comprising both Laplace noise and BPSK-modulated signals under various signal-to-noise ratio (SNR) conditions. The results indicate that this method has a better detection probability under low SNR conditions than the standalone LSTM and traditional spectrum sensing methods.