Research On Computation Offloading Strategy Based On Mobile Edge Computing

The rapid development of Internet of Things technology and wireless communication technology has promoted the process of information society.In the context of the Internet of Everything,smart terminal devices connected to the network are rapidly popularizing,and various applications are becoming more and more complex.Due to the small battery capacity and insufficient computing power of terminal devices such as mobile phones and tablets,it is difficult to handle these huge computing tasks.The traditional cloud computing method has sufficient computing resources,but it is far away from the user terminal equipment,which will cause a large time loss in the process of transmitting data,and it is difficult to meet the delaysensitive tasks such as automatic driving,virtual reality,online games and other needs.As an emerging computing paradigm,Mobile Edge Computing(MEC)provides computing,storage and control functions of cloud computing at the edge of the network by deploying servers on infrastructure such as base stations.Therefore,MEC is closer to the user terminal and can provide users with a faster and safer service.The computation offloading technology in MEC can help users decide how to execute tasks,that is,whether to choose to offload and how much to offload,so as to adapt to the requirements of different applications.Computation offloading affects user experience and has become a research hotspot in MEC networks.Therefore,this paper mainly conduct research on computation offloading strategies in MEC systems.For the MEC scenario of single server and multiple users,two computation offloading methods are proposed.1、In the case of overall task offloading,this paper provides a computation offloading strategy based on genetic algorithm.First construct the MEC network scenario and mathematically model the optimization objective.Then the objective function in the model is transformed into the fitness function of the genetic algorithm for screening excellent individuals.The algorithm finally reaches the convergence state through continuous iteration,that is,the computation offloading strategy which minimizes the total consumption of the system is found.Experimental data show that the total consumption of this algorithm is lower than that of the baseline algorithm,and it also has good performance in optimizing delay or energy consumption alone.2、In view of the partial offloading of tasks,this paper provides a computation offloading strategy based on reinforcement learning.First,under the constraints of various constraints,this paper constructs a mathematical model with the weighted sum of delay and energy consumption as the optimization goal,and then gives the specific content corresponding to the state,action,and reward in reinforcement learning.Finally,the Q-learning algorithm is used to solve the problem of computation offloading and resource allocation.Experimental data show that the proposed strategy can further decrease the total consumption of the system under different parameters.