Abstract:Cooperative spectrum sensing is a key technology to solve the problems of multipath fading, shadowing effect and receiver uncertainty of a single cognitive radio (CR) node. First, four common cooperative spectrum sensing models are sorted out, and the advantages and disadvantages of the four models as well as their usage scenarios are compared and analyzed in detail. Then, the research progress of cooperative spectrum sensing based on reinforcement learning is reviewed, and the advantages and disadvantages of several reinforcement learning algorithms as well as their applicability scenarios are compared and analyzed in detail, and it is found that there is still a great potential to reduce the synchronization and communication overheads of cooperative sensing. Finally, the current challenges and future research directions in cooperative spectrum sensing techniques based on reinforcement learning are summarized, and the use of a multi-agent deep deterministic policy gradient (MADDPG) algorithm is proposed to solve the mobility synchronization and communication overhead problems of the cognitive radio network in cooperative sensing technology problem.