| Mobile Edge Computing(MEC)can effectively compensate for the shortcomings of cloud computing,but still cannot cope with emergency situations of natural disaster rescue or incomplete infrastructure.The most researched method to solve this problem is UAV-assisted MEC technology.In this paper,we further expand the multi-UAVsassisted MEC system to a multi-mobile-platforms-assisted MEC system by expanding the assistive devices from UAVs to all devices that meet the definition of mobileplatform,and by exploiting the commonality between different mobile-platforms to find a unified deployment and offloading method.In a multi-mobile-platforms-assisted MEC system,multiple mobile platforms configured with communication modules and computing modules are introduced,and the computing module is referred to as MP-MEC server.Also,the devices in the system that can provide computational resources other than the MP-MEC server are called BSMEC server.In the multi-mobile-platforms-assisted MEC scenario constructed in this paper,there is one BS-MEC server,multiple MP-MEC servers,and multiple end users.This paper focuses on the computational offloading problem in the multi-mobileplatforms-assisted MEC system,aiming to find an intelligent offloading scheme to allocate the computational resources of multiple edge servers through a reasonable offloading strategy,improve the utilization of resources,and help the end-users in the system to complete the computational offloading operations efficiently.For each enduser,three data processing methods are available:(1)Local execution;(2)Offloading to the BS-MEC server;(3)Offloading to one of the MP-MEC servers.The main research idea of this paper is to implement two rounds of decision making.The first-round decision is completed through a pre-offloading process to determine the offloading destination of computational tasks;the second-round decision is performed through a resource pricing offloading incentive mechanism to determine the amount of data to be offloaded for each computational task.The main research components are as follows.(1)Pre-offloading operation is performed for each task in the system,and the computational tasks are assumed to be indivisible tasks for complete offloading.First,the total system utility function is constructed using three factors: energy cost,time cost and overall satisfaction of user.Then,the task offloading destination decision problem in the multi-mobile-platforms-assisted MEC system is modeled as a system total utility optimization problem under the constraints of resource competition and offloading decision,and three intelligent offloading algorithms,RSTO algorithm,QLO algorithm and DQNO algorithm,are designed to solve the optimization problem.Finally,it is demonstrated through simulation experiments that the DQNO algorithm and the RTSO algorithm can be used as the optimal and suboptimal algorithms for task offloading destination decisions,respectively.(2)After determining the offloading destination of computing tasks,an incentive mechanism is introduced to study the data offloading problem between a single edge server and its corresponding computing tasks.First,the edge server utility model is constructed considering the reward and energy consumption factors,and the user-side utility model is constructed considering the user satisfaction,energy consumption and offloading cost factors.Then,the interaction process between the edge server and the end user is modeled as a Stackelberg game problem,and the existence of a Nash equilibrium in the Stackelberg game is demonstrated through theoretical and simulation experiments.Finally,the MRIG algorithm is designed to solve the Nash equilibrium solution of the Stackelberg game,and the effectiveness and superiority of the MRIG algorithm are proved by simulation experiments. |
PDF Full Text Request