Research on unlicensed spectrum access mechanism based on reinforcement learning in LAA/WLAN coexisting network

摘要

The efficient and fair coexistence of Licensed-Assisted-Access (LAA) and WLAN has become an urgent problem to be solved. According to the heat problems above, this thesis has conducted an in-depth research on access mechanism of LAA and WLAN systems in unlicensed frequency bands. The study firstly introduced an access mechanism based on reinforcement learning after making the summery of the existing coexistence mechanisms. This mechanism takes Listen Before Talk mechanism as a technical basis to have adaptive adjustment of the channel access probability, thus to increase the system capacity. Meanwhile, the system's throughput is processed using a proportional fair utility function, and utility value is considered as the return value of the system to ensure fairness between the LAA and the WLAN. Simulation results indicate that the algorithm proposed in this study can effectively maximize the system throughput. In the case of different LAA and WLAN packet lengths, algorithm proposed can still guarantee fairness among users in coexistence system.