| As the sixth-generation mobile communication system gradually enters reality from conception,massive data-intensive applications and wireless communication devices are in urgent need of more bandwidth resources,so the contradiction between supply and demand of spectrum use is becoming more acute.At present,there are two ways to solve the shortage of spectrum resources.One is to expand to a higher white space frequency band,and the other is to improve the utilization efficiency of the existing frequency band.However,high frequency band has the characteristics of short wavelength,large transmission energy loss,and higher requirements on the precision of hardware equipment.Therefore,maximizing the utilization efficiency of existing frequency band through dynamic spectrum access technology is still one of the most concerned problems at present.But at present,dynamic spectrum access technology still has great interference to the main user,frequent spectrum switching leads to the decline of cognitive user communication quality.Therefore,in order to solve the above problems,this paper firstly studies the spectrum occupancy rule of the primary user,and proposes a new hybrid spectrum prediction algorithm based on deep learning by combining the spectrum state rule with the prediction of the available channel duration,so as to achieve the goal of improving the prediction accuracy and reducing the conflict with the primary user.Secondly,the dynamic spectrum access process under the distributed non-cooperative system model is studied.Aiming at the problems that users need to switch channels frequently during the access process and their communication service quality deteriorates,the dynamic spectrum access algorithm is optimized by combining the prediction knowledge,and the prediction knowledge is taken as the state space input of the decision network.At the same time,considering the influence factors of cognitive users in the process of non-cooperative access to the channel,simulation experiments are carried out under saturated services.The results show that cognitive users can effectively reduce the channel switching delay and communication interruption probability,and the optimized algorithm has less interference to the main user.Finally,this paper analyzes the problems of inaccurate perception and communication interference caused by the dynamic spectrum access process considering the user’s spatial distance.In this paper,the echo state network structure is introduced to remember the spectrum access action,and the historical spectrum state is learned to avoid the collision between users caused by perceptual errors.In addition,the reward function of the decision algorithm is improved,and the communication interference caused by cognitive users to the main users is limited by introducing the protection distance of the main user space.Experiments show that the improved algorithm can effectively reduce the conflicts among users and improve the efficiency of spectrum access. |
