Deep Reinforcement Learning-Based RIS-Assisted Cooperative Spectrum Sensing in Cognitive Radio Network

Cognitive radio network (CRN) is considered to be an effective means of improving spectrum utilization. As a crucial technology in CRN, spectrum sensing detects spectrum holes to achieve efficient frequency band utilization without interfering with licensed users. However, practical scenarios face challenges like wireless channel impairments, such as shadowing and path loss over long distance, which degrade secondary user (SU) reception of primary user (PU) signals. Reconfigurable intelligent surfaces (RIS) offer a promising solution by dynamically adjusting channel conditions to enhance wireless communication efficiency. This paper introduces a RIS-assisted cooperative spectrum sensing scheme to bolster PU signal reception by SUs, thereby improving spectrum sensing reliability. Leveraging deep reinforcement learning, our proposed approach optimizes cooperative spectrum sensing by efficiently coordinating SU actions. Through numerical simulations, we demonstrate the effectiveness of our method, which outperforms existing approaches in terms of detection performance. In particular, the number of SUs required by our proposed efficient spectrum sensing algorithm is fewer than the total number of cooperative spectrum sensing.