Research On Resource Optimization Methods For Minimizing Delay In Mobile Edge Networks

In recent years,with the rapid development of the Internet and the increasing popularity of intelligent terminals,new delay sensitive applications such as face recognition,augmented reality,virtual reality,interactive games and real-time holographic projection have been emerging.Mobile Cloud Computaion(MCC)is no longer capable of meeting the latency requirements for such applications and has resulted in revolusion of mobile network architecture.Different from MCC,Mobile Edge Computation(MEC)can provide low-latency computing services for nearby User Equipment(UE)by deploying MEC servers on the edge of the network.However,MEC server needs to use limited resources to provide services for the UEs.Thus how to optimize the allocation of resources has become a key issue in mobile edge networks.Meanwhile,the explosive growth of offloading traffic also brings huge pressure to the transmission network,and how to reduce the amount of data transmission has become a research hostpot in mobile edge networks.Based on the above analysis,this thesis introduces data compression technology into MEC,and proposes the delay minimizing resource optimization algorithms in mobile edge networks.The main contributions of this thesis are as follows:(1)For the single MEC server multi-user scenario,this thesis proposes a resource optimization algorithm based on Particle Swarm Optimization(PSO).Under the maximum energy consumption constraint of UE,the UE's task offloading ratio,the compressed ratio,the tansmit power and the computation resource allocation of the MEC server are jointly optimized,with the goal of minimizing the total task execution delay of UEs.Simulation results show that the proposed algorithm can significantly reduce the total task execution delay of UEs.In addition,compared with traditional PSO,random weight PSO and linear decreasing weight PSO,the proposed algorithm has stronger global search ability and is not easy to fall into the local optimization.The simulation results also show that with the increase of task data and task intensity,the proposed algorithm is better in reducing the total UE task execution delay compared to the algorithm without data compression.(2)For the multi-MEC server multi-user scenario,this thesis proposes a resource optimization algorithm based on an improved Genetic Algorithm(GA).Different from the first scenario,this scenario considers the joint optimization of computing resources and communication resources to minimize the total task execution delay of UEs.In addition,in order to improve the global search ability of traditional GA,this thesis improves the traditional GA in the following two aspects: a)The population learning idea in PSO is used to ensure that the optimal individuals of the population can be selected in GA selection.b)After GA crossover and mutation operations,fitness values are evaluated for the population,so as to retain the global optimal solution generated in crossover operations.Simulation results show that,with the increase of the numbers of UEs and MEC servers,the performance of the proposed algorithm is better in reducing the total UE task execution delay compared with the nearest offloading method and the random offloading method.