On-Demand Joint Resource Management Technology For 6G

To provide users with on-demand high-capacity services,the infrastructure of the 6th generation wireless communication networks(6G)will be extended from the ground to the air and arise space-air-terrestrial integrated relationship.Moreover,the 6G will introduce multi-dimensional resources.Joint management of communication,caching and computing resources can improve the service capability without increasing the network deployment cost.Hence,it is critical to design the deployment as well as scheduling policies of multi-dimensional resources and further realize on-demand joint resource management based on the characteristics of network and service.There are some challenges of the resource management technology for 6G.First,the dense deployment of ground base stations(GBSs)in 6G makes the coupling relationship of resources in adjacent GBSs more complex.Meanwhile,the uneven distribution of network resources is becoming more prominent.Then,air base stations(ABSs),as an important part of 6G,can expend the coverage area of network and realize on-demand services.However,the joint resource management of ABSs is time-varying due to the mobility of the ABS,which makes the service continuity difficult to guarantee.Final,there is strong inter-cell interference(ICI)when GBSs and ABSs cooperatively serve users.Moreover,it is difficult to match the resources of access and backhaul of ABSs,which makes joint resource management in ABSs and GBSs challenging.How to use edge computing to simplify the resource management and realize real-time on-demand matching of services and resources is critical to joint resource management of ABSs.Thus,this dissertation studies the joint resource management technology for 6G to match resources and service requirements and further improve the network service capability.The main contributions of this dissertation can be summarized as follows:1.The uneven distribution of resources makes the network congestion and the decline of throughput capacity.To solve it,a social-aware content caching and delivery policy is proposed,which jointly manages the communication and caching resources and exactly match the traffic distribution with the structure of networks.First,the effective betweenness(EB)based analysis method of traffic distribution is proposed.The results show that it is necessary to preferentially cache high-popularity contents in users with low sociality and high sharing willingness to maximize the throughput capacity without congestion.Aided by it,a symbolic geometric programming-based algorithm framework of content caching and delivering is developed.Then,the optimization method and iterative method are explored to design the content caching and delivery strategy and improve the throughput capacity.Simulation results show that the proposed algorithm can jointly manage communication and caching resources according to network structure and node characteristics,and further improve throughput capacity without congestion.2.The service continuity of ABSs is difficult to guarantee due to the mobility of ABS.To address it,a policy of ABS deployment and content delivery with high energy efficiency is proposed.The policy can generate the dynamic management strategy of communication and caching resource aided by the edge computing,and further guarantee service continuity under the limited energy and backhaul capacity.First,a trajectory planning problem is formulated to maximize the time of continue coverage considering the content caching and charging process of ABSs.Since only the local information can be obtained by ABSs,the trajectory planning problem is further formulated as two coupled multi-agent stochastic games,whose equilibrium solutions constitute the optimal trajectory planning.Then,a distributed RL framework is designed to obtain the equilibrium solutions of the two games above.The proposed framework can decouple the training process of the two types of agents,thereby accelerating the convergence rate.The theoretical analysis proves that the proposed distributed RL algorithm can converge to the optimal solution of the Bellman equation.Simulation results show that the proposed algorithm can optimize the trajectory of ABS and charging scheduling to guarantee the service continuity with high energy efficiency under the unknown environment information.3.The mismatch of resources of access and backhaul will greatly limit the downlink rate of the ABS-aided cellular system.To solve it,a air-ground integrated resource management method is proposed.The method can jointly manage resources of access and backhaul aided by the center-edge computing,and further achieve non-interference access and high-capacity backhaul.First,a hybrid coordinated multi-point(CoMP)based transmission structure is proposed,where two different coordinated multi-point(CoMP)modes are applied in access links and backhaul links of ABSs according to the capacity of data sharing of BSs.Then,a robust optimization framework of network deployment and resource management is designed considering the randomness of air-ground channels to optimize the resource allocation policy of access and backhaul links.The initial multi-timescale problem is divided into multiple single-timescale subproblems via sample average approximation.Furthermore,the alternating direction method of multipliers(ADMM)is explored to restore the solution of the initial problem from that of subproblems.For each subproblem,both the monotonic optimization based optimal algorithm and the cascade of matching game and RL-based online algorithm are designed.Simulation results show that the proposed algorithm can avoid the strong interference caused by ABS and improve service capacity via on-demand matching resources of access and backhaul links.