Research On Key Technologies Of Resource Allocation And Cache Optimization In Dynamic Access Network

In dynamic access network,the network entities such as member nodes,access points(APs)generally have mobility,which causes frequent changes in member nodes associated with network.Moreover,the links between the network entities may be under unstable channel conditions.Hence,it is essential to allocate recource of networks against such dynamics.This thesis studies the resource allocation and caching optimization of dynamic access network,mainly focuses on three different network scenarios: wireless patrol network,User-centric Ultra Dense Network(UUDN)based on dynamic network architecture(DNA)and device to device(D2D)caching networks with multiple mobile helpers.The main contributions are as follows:Firstly,the wirelss patrol network is a kind of dynamic access network since the location of a hybrid access point(HAP)vary in time.we study the resource allocation of wireless patrol network cooperated by backscatter communication with finite alphabet inputs.To maximize the network throughput,we propose a two dimension method to jointly allocate power and time.The simulation results show that integrating backscatter communication into the traditional wireless patrol network can also improve the network performance.Secondly,we study the resource allocation of UUDN,both users and APs work in full-duplex mode,user in the network can switch to an access point(AP)dynamically for other users to access.Considering the self-interference and the interference between AP and user.This thesis introduce a joint scheme of user access,subchannel allocation and power control for UUDN.the problems of sum-rate maximization and total power minimization are studied,respectively.we propose the MLFP-based power allocation(MAPEL)and the improved Particle Swarm Optimization(PSO)algorithm to solve the sum rate maximization problem,respectively.The power minimization problem is transformed by averaging the total user rate in each subchannel of the user,transformed into a convex optimization by time-sharing method and then solved by Lagrange duality.The simulation demonstrates the outperformance of UUDN over CUDN and the advantages of FD over HD in terms of spectrum efficiency and power consumption.Finally,we investigate an optimal caching scheme for D2 D networks with multiplemobile helpers with a large cache size.we proposed a algorithm based on the delay-aware caching(DACA)algorithm to obtain optimal caching scheme for helper at a certain position,then adopt particle swarm optimization(PSO)algorithm to get the optimal position of helpers to minimize the average delay and power consumption of the network.The simulation results show that the helper can reduce the system delay and energy consumption,and multiple helpers can better improve the system performance to some extent.