Deep Learning for Cooperative Spectrum Sensing in Cognitive Radio
Zhibo Chen, Daoxing Guo, Jie Zhang · 2020
In this paper, we propose a collaborative spectrum sensing algorithm based on deep learning. Firstly, we use distributed secondary users (SUs) to receive sensing samples, which effectively avoids the channel's fading and shadow effects. Then, we propose a detection framework based on deep learning and select the sample covariance matrix as the test statistic. To realize the detection framework, we put the sample covariance matrix into a convolutional neural network (CNN) and propose a new cooperative spectrum sensing algorithm named CSS-CNN algorithm. Besides, theoretical analysis for the proposed algorithms is given, and find the associated threshold. We also analyze the robustness of SNR and noise uncertainty for the CSS-CNN algorithm. Notably, the proposed algorithm requires neither any information about the signal nor noise power. Simulation results demonstrate that the proposed algorithm outperforms other popular sensing methods based on deep learning and conventional sensing methods.