Cooperative Sensing Based on Coverage Prediction in Heterogeneous Spectrum Availability

Abstract

This letter proposes a novel cooperative spectrum sensing (CSS) scheme based on primary users (PUs) coverage prediction for cognitive radio networks in heterogeneous spectrum availability (HetSA) environments. To achieve accurate sensing, a conditional generative adversarial network (cGAN) is employed to learn and construct the radio map, based on limited sensing data from secondary users (SUs). Subsequently, an improved density peak clustering (DPC) algorithm is used to identify the density distribution of PU signal strength and accurately predict PU coverage areas, exhibiting robustness against channel fading and noise. Finally, a radio map coverage prediction based CSS (RMCP-CSS) detector is proposed, which efficiently fuses predicted PU coverage for enhanced spectrum sensing. Simulation results demonstrate that the proposed RMCP-CSS achieves superior detection performance compared to state-of-the-art CSS algorithms, making it well-suited for small-scale, dense sensor deployments in HetSA scenarios.