Computation Offloading Strategy Based On Multi-strategy Adaptive Bat Algorithm In Mobile Edge Computing

In recent years,with due to the rapid development of cloud computing,Internet of Things and other technologies,more and more new network applications with computing intensive and time-delay sensitive characteristics is appearing in users’ vision.To overcome the limitations of Mobile devices for new network applications in the workload,Mobile Edge Computing(MEC)can effectively solve the problems related to the energy and computing capacity of Io T devices,and then becomes the key technology of the Next Generation Network.Task offloading and resource allocation schemes in edge computing system are widely concerned by industry and academia.In the current research,computing offloading technology is used to offload new network applications from mobile user devices to edge servers.However,there are some shortcomings in this process,such as high system cost of unloading scheme,in discriminating the restriction of delay,and the balance between delay and energy consumption is not considered.Therefore,this paper proposes a multi-task computational unloading model and an adaptive discrete bat algorithm to solve the decision vector of the model to further improve the computational unloading function.The main research work of this paper includes the following two aspects:Firstly,aiming at the problem that task running time and task energy consumption need to be optimized and limited in moving edge computing,this paper presents a multi-task computing unloading model that comprehensively measures service delay and energy consumption ratio under the condition of energy consumption constraint.Specifically,a mixed integer nonlinear programming problem is constructed for moving edge computation,and it is solved by analyzing the convexity and constraints of cost function.Second,a multi-strategy adaptive bat algorithm(MABA)is proposed to solve the decision vector of unloading after model optimization.In the first place,bat algorithm(BA)is modified to run in discrete case.Then,in the global search mode of bat algorithm,the adaptive search mode,the multi-strategy design and the combination of algorithm and random flight strategy are adopted to make up for the defect of early convergence,to strengthen its global search ability in discrete space and avoid falling into local optimal solution.Finally,the local search method of bat algorithm is designed by adaptive and multi-strategy to enhance its local search ability,so that the algorithm can carry out effective local optimal search in discrete environment.Through simulation experiments,the MABA algorithm is compared and analyzed with artificial fish swarm algorithm(AFSA),bacterial foraging algorithm(BFA),genetic algorithm(GA),particle swarm optimization algorithm(PSO),local computing(LC)and random offloading(RO)in various dimensions.Experimental results show that the MABA is superior to the benchmark algorithm in all indicators,MABA algorithm can significantly reduce the total cost of unloading calculation,which can be better applied to the moving edge calculation.