Research On Offloading Decision And Resource Allocation Strategy For Mobile Edge Computing With Computing Resource-Constrained

With the rapid development of 5G technology and relevant computation-intensive applications such as VR and AR,the computing power and battery energy of user devices are facing great challenges.Mobile Edge Computing(MEC),as the network architecture of 5G evolution,can effectively solve the contradiction of computing power,delay and energy consumption between user equipment and application by providing computing and storage capacity at the network edge.However,with the growth of users and cost constraints,the limitation of MEC server's computing resources has become increasingly prominent.Therefore,how to design a reasonable offloading decision and resource allocation strategy in MEC servers with limited computing resources is one of the problems that need to be solved urgently.For this reason,in the MEC offloading environment,this article increases the constraints in the real offloading situation such as the limited computing resources of the MEC server and the limited task completion time,and designs effective offloading decision and resource allocation strategy to shorten the system delay and reduce the terminal energy consumption.The specific research work is summarized as follows:(1)Aiming at the multi-user single-server offloading scenario with limited computing resources of MEC server,partially improved the genetic algorithm based on the elite selection strategy,and a joint optimization strategy improve-eGA of offloading decision and resource allocation is proposed to minimize the total system cost(weighted sum of execution delay and energy consumption).Through simulation experiment,compared with classical local execution algorithm,improve-EGA can reduce the total system cost by 25%.(2)Aiming at the multi-user and multi-server offloading scenario with limited computing resources of MEC servers,increase the task completion time constraints,designed a new type of objective function and partially improved the Nature DQN algorithm in deep reinforcement learning,and proposed a joint optimization strategy Based DQN of offloading decision and resource allocation in order to shorten the task completion time and reduce the terminal energy consumption while meeting the delay constraint.Through simulation experiment,compared with classical local execution algorithm,the Based DQN under the new objective function can shorten the task completion time by 33.51%under the delay constraint and reduce the terminal energy consumption by 51.51%.