Resource Allocation Algorithm Based On Stochastic Optimization For 5G Network Slicing

As one of the key technologies of 5G,network slicing divides the actual network into virtual networks corresponding to different application scenarios by dividing the network resources and functions.Resource allocation of network slicing is the primary consideration for flexible on-demand networking.Efficient resource allocation algorithms can improve network resource utilization,ensure user's quality of service,and increase network performance.Therefore,this thesis focuses on the resource allocation of 5G network slicing based on the stochastic optimization theory.The main research contents and work are summarized as follows:1.To satisfy the diversity of requirements for different network slices and realize dynamic allocation of wireless virtual resource,this thesis proposes an algorithm for network slicing joint user association and power allocation in cloud radio access network based on non-orthogonal multiple access.Firstly,by considering imperfect channel state information,this thesis develops a joint user association and power allocation algorithm to maximize the average total throughput in cloud radio access network with the constraints of slice and user minimum required rate,outage probability and fronthaul capacity limits.Secondly,this thesis designs a dynamic joint user association and power allocation algorithm by transforming the probabilistic mixed optimization problem into a non-probabilistic optimization problem and using Lyapunov optimization.Finally,for user association problem,this thesis proposes a greedy algorithm to find a feasible suboptimal solution;and transforms the power allocation problem into a convex optimization problem by using successive convex approximation,then exploits a dual decomposition approach to obtain a power allocation strategy.The simulation results show that the proposed algorithm can effectively improve the average total throughput of network slicing while guaranteeing the requirement of each slice and user.2.Aiming at the unreasonable allocation of time-frequency resources caused by the uncertainty of content request and the shortage of spectrum resources in network slicing,this thesis proposes an algorithm joint time-frequency resource allocation and content caching in fog radio access networks.Firstly,according to the dynamic randomness of different network slice content requests and the quality of service of the fronthaul link and the wireless access link,a constrained markov decision process stochastic optimization model that maximizes the long-term average system utility is established under the constraint of the content servicedelay and the network slice on the minimum demand of time-frequency resources.Then because it is difficult to obtain the network state transition probability,the post-decision state is introduced to establish the dynamic programming equation based on the post-decision state value function.Finally,an online learning algorithm based on stochastic gradient method is proposed so as to obtain the time-frequency resource allocation and content caching algorithm.The simulation results show that the proposed algorithm can better balance the content cache decision and resource allocation decision while ensuring the network slice content service delay.