With the development of wireless technology,the spectrum resource is becoming more and more scare.The traditional fixed spectrum allocation strategy severely limits the ability to access,resulting in serious waste of spectrum resources.As the primary users occupy spectrum,it will set restriction for secondary(or cognitive)users.The amount and position of the spectrum that the secondary users can use will change over the time.The problem is actually how to allocate spectrum under certain restriction.Cognitive radio technology can solve the problem efficiently by integrating spare spectrum,which can improve the spectral efficiency.Q learning algorithm has been used in a variety of situations.It can also be applied to provide a spectrum resource optimization method to enhance the spectral efficiency.This passage introduces and analyzes the model of dynamic spectrum access in cognitive wireless network and the theory of Q learning firstly.Then the algorithm and the framework of Q learning algorithm is given.Under the architecture of cognitive radio network,a modified Q learning-based dynamic spectrum access optimization strategy is proposed.By channel request and node-control,the strategy can achieve dynamic spectrum access.Simulation and evaluation show the proposed scheme can improve the overall network performance and enhance the fairness of access for cognitive user,as well as increase the capacity of cognitive systems.