Research On 5G Heterogeneous Ultra Dense Network Selection Algorithm

As the global communication industry continues to develop and construct,the 5th Generation Communication technology(5G)has been commercially launched worldwide in2020.At the same time,with the widespread adoption and application of intelligent devices such as smart portable terminals and wearable communication devices,users’ communication services have increasingly higher requirements for Quality of Service(Qo S).As a result of differences in performance and communication technologies among various wireless networks,any single wireless network cannot meet the increasing and diversified Qo S demands of users.Therefore,in the 5G era,various networks have been integrated to form a Heterogeneous Ultra Dense Network(HUDN),which will become a future development trend of wireless networks.There are multiple networks in HUDN,to help users select appropriate networks according to different service requirements,an efficient and reliable network selection algorithm needs to be designed.This dissertation comprehensively analyzes and sorts out the existing network selection algorithms,presents the fundamental concepts and theories of each algorithm,assesses their strengths and weakness,and propose the network selection algorithm based on Dueling-DDQN and the network selection algorithm based on fuzzy-gray ideal solution.The main contents of this dissertation are as follows:(1)Existing network selection algorithms consider few Qo S indexes and fail to effectively consider users’ network selection preferences,and existing network selection algorithms fail to effectively distinguish different business types of users.This dissertation suggests a novel algorithm about network selection based on Dueling-DDQN,models the network selection problem as Deep Reinforcement Learning(DRL)methods,considers multiple Qo S indexes of user services,and introduces analytic hierarchy process and standard deviation method to obtain the preference weights of different user services for different attributes of network end.Based on different business types initiated by users,the reward functions of different business selection networks are designed by using utility functions,which effectively distinguish different business types of users.Simulation results show that the proposed algorithm can improve the benefits of network selection and the resource utilization efficiency of the network side.(2)In order to solve the problems of frequent network switching caused by mobile users,high handover delay and low algorithm identification in existing network selection algorithms,a network selection algorithm based on fuzzy-gray ideal solution is proposed.By calculating the signal strength of different networks,networks higher than the threshold value of signal strength are selected to form alternative network sets.In the selection of parameters at the network end participating in decision-making,different types of services initiated by users are selected according to the Qo S requirements of different services.By reducing the parameters involved in the network selection decision,the calculation amount of the algorithm is reduced and the user handover delay is reduced.Finally,the membership function of different network attribute parameters is designed,and the fuzzy decision matrix is constructed.Using improved gray ideal algorithm to acquire the score of each alternative network,and generate the network selection output by judging the threshold value.The simulation results indicate the algorithm mentioned in this dissertation effectively decreases the handover delay and ameliorated the accuracy user’s network selection.