Research On Radio Resource Management For 5G Heterogeneous Ultra Dense Network

With the development and commercialization of the Fifth Generation(5G),the access technology of next-generation wireless networks has become more complex.5G is converging with existing communication technologies to become Heterogeneous Ultra Dense Network(HUDN),which supports real-time communication at terminals.In addition,Non-Orthogonal Multiple Access(NOMA)technology,a candidate for 5G multiple access,can allocate the same carrier resource to multiple users through the power domain or code domain,improving spectrum efficiency and user access.Therefore,HUDN and NOMA technologies are considered as two key technologies to increase system capacity in 5G communications.However,in the HUDN environment,their dynamics are greatly increased,which in turn affects the system performance and user’s Quality of Experience(Qo E)of service.In this dissertation,we propose a wireless resource allocation algorithm based on an improved Deep Q-Network(DQN)and a competing Double Deep Q-Network(Dueling DDQN)with the objective of improving the system performance and user Qo E.algorithm.The proposed two algorithms are optimized at two different levels,namely maximizing user Qo E scores and system energy efficiency(EE),in order to improve the overall system performance and resource utilization while satisfying the user Qo E:(1)To study the allocation of NOMA wireless resources under HUDN in view of the complex and time-varying characteristics of the HUDN environment.To improve the service Qo E of network users,a joint optimization Qo E model based on Mean Opinion Score(MOS)is developed to maximize the cumulative sum of MOS of all network users,and the wireless network resource allocation problem is modelled as a Markov Decision Process(MDP).The proposed algorithm is based on Double Deep Q-Network(DDQN),and the details of the proposed algorithm are presented.The proposed algorithm is verified to be effective in improving the Qo E of users and enhancing the carrying capacity and performance of the communication network through simulation.(2)To address the problems of low system EE,serious cell interference and unsatisfiable user Qo E in the existing NOMA-based 5G HUDN research.Firstly,the downlink NOMA twolayer heterogeneous network communication model and HUDN energy-efficient resource allocation optimization function are established,and the complexity of the optimization problem is thus reduced by solving two sub-problems of sub-channel allocation and power resource allocation step by step;then cell interference is reduced through base station clustering;The simulation results show that the proposed algorithm can effectively improve the system EE,reduce cell interference and improve the utilisation of spectrum resources and energy efficiency resources.