A Dynamic Spectrum Access Method Based on Q-Learning

Dynamic spectrum access is an important step in the spectrum resource sharing process. However, in the case of unknown spectrum access environment and dynamic model, it is difficult to achieve global optimal resource allocation. Considering the signal-to-noise ratio (SINR) limits of the primary network (PN) and the secondary network (SN), this paper presents a dynamic spectrum access method based on the relationship between the transmission rate of a secondary user (SU) and the relative SINR, which enables the SU to obtain and update the environment information through the Q-function, adopt an optimal strategy to maximize the network performance and utilize user collaborative learning mechanism to overcome local optimum problems. The simulation results show that the number of iterations required for the proposed algorithm to achieve convergence is reduced by up to 65%, the average performance index of the system is improved by up to 35% and the proposed algorithm can also ensure the fairness among SUs compared with traditional access method.