Resource Scheduling In Mobile Edge Computing For Internet Of Vehicles

With the popularity of the Internet of Vehicles,a number of vehicle applications are emerging.Completing the tasks of these applications requires the processing of video,image,voice data from the sensors of its own and other surrounding vehicles.The amount of data is quite large,and the processing of these data requires strong computational capabilities and limited vehicle computing capabilities require that such computationally intensive tasks should be uploaded to the data center for processing.The data size of uploaded tasks is quite large and the Qos required by vehicle applications is demanding in terms with latency and reliability.MEC technology is a good choice for this kind of "high bandwidth,low latency,high reliability" scenario.After the combination of LTE architecture and the MEC technology,data centers for processing user requests will be deployed around the small-cell base station(SBS),the macro-cell base station(MBS),as well as the Internet.So users can access either SBS or MBS to offload tasks.The user's access selection,task unloading destination selection,and wireless and computing resource allocation scheme during the task unloading process all affect the efficiency of system-level task execution.Our paper focuses on solving the problem of tasks offloading decision and the jointly resource allocation of wireless time slot and data center CPU cycles.Our goal is to minimize the average tasks completion time,under under the circumstance of ensuring the requirements of each task,thus improve the task execution efficiency of the entire system.This paper proposes a Jointly Task Offloading and Resource Allocation Optimization(TORA)algorithm,which decomposes the task unloading decision and resource allocation scheme into two sub-problems and solves two optimal sub-problems respectively.It also demonstrates the feasibility of the algorithm,and simply explains the existence of the lower bound of the original optimization problem.Finally,simulation experiments are conducted to analyze the influence of parameters such as number of users,amount of user task upload data,number of CPU cycles required for user tasks,maximum task completion time,etc.on the efficiency of tasks execution in the whole system.We compared the simutation results of TORA algorithm with other two algorithms,as well as the lower bounds of the original optimization problem.