中圖分類號：TN925 文獻標志碼：A DOI: 10.16157/j.issn.0258-7998.234333
中文引用格式： 李新玉，趙知勁. 基于深度學(xué)習(xí)的動態(tài)主用戶頻譜感知算法[J]. 電子技術(shù)應(yīng)用，2024，50(1)：60-65.
英文引用格式： Li Xinyu，Zhao Zhijin. Dynamic primary user spectrum sensing algorithm based on deep learning[J]. Application of Electronic Technique，2024，50(1)：60-65.

Dynamic primary user spectrum sensing algorithm based on deep learning

Abstract： In actual spectrum sensing scenarios, the primary user may arrive or leave randomly, and when the primary user state changes dynamically during real-time spectrum sensing, the performance of the existing static spectrum sensing algorithm deteriorates sharply. For this situation, this paper propose a dynamic primary user spectrum sensing algorithm based on the residual shrinkage and attention mechanism. During the spectrum-sensing interval, the time when the primary user randomly arrives or leaves randomly follows a uniform distribution. The “deep residual shrinkage network (DRSN)” is used to extract dynamic primary user features and filter out redundant noise features. The “coordination attention module (CAM)” is used to improve the ability of the model to express the features of the dynamic primary user. Simulation results show that the proposed algorithm performs are better than ResNet algorithm, CBAM_IQ algorithm and CBAM_Energy algorithm. The proposed algorithm can maintain a high detection probability for the primary users who randomly arrive or leave following different distributions.

Key words : cognitive radio；spectrum sensing；dynamic primary user；deep residual contraction network；coordinated attention mechanism