Research On Differential And Game Evolution Methods For Task Offloading In Mobile Edge Computing

The development of 5G / 6G provides technical support for the connection of various edge and terminal devices and high-speed data transmission.Mobile communication is no longer limited to providing voice services.Emerging applications such as online games,augmented reality,driverless,telemedicine diagnosis came into being.However,these emerging intelligent applications are often computation-intensive and delay-sensitive,while most terminal devices are limited in computing resources.There is a contradiction between the stringent requirements of emerging applications on computing resources and the processing capacity of terminal devices.Although traditional cloud computing technology can use the powerful computing power of remote cloud data center to provide services for users,its centralized processing method will lead to network congestion and increase response delay.Mobile edge computing sinks the cloud services in cloud computing technology to devices at the edge of the network,which has important potential to meet the real-time requirements of emerging applications,and has attracted much attention in recent years.As mobile edge computing nodes are dispersed and heterogeneous,with the increasing number of users and tasks,reasonable task offloading and resource allocation strategy is the key to achieve efficient computing.Therefore,this paper focuses on the task offloading and resource allocation optimization problem in mobile edge computing.The main research contents are as follows:(1)Under the single-layer MEC computing architecture,each user’s computation task has a strict completion time limit.The weighted sum of delay and energy consumption is taken as the task computing cost,and the overall goal is to minimize the total computing cost of all users’ tasks,forming a joint optimization problem of task unloading and resource allocation.To solve this optimization problem,a CORA＿IDE global optimization algorithm based on differential evolution algorithm is proposed and its good performance is verified by experiments.Compared with the three benchmark methods,the performance of CORA＿IDE algorithm is improved by76% to 93%,which effectively achieves the purpose of tradeoff optimization between task calculation delay and energy consumption.(2)Under the two-layer MEC computing architecture based on edge-edge collaboration,the edge server deployed in the small base station and the edge data center deployed in the macro base station cooperate closely to provide users with high-quality computing services.Each mobile user has multiple independent subtasks with limited completion time,these subtasks can be processed by local computation and offloading computation simultaneously.The weighted sum of delay and energy consumption is taken as the task computing cost,and the overall goal is to minimize the sum of the computing costs of all subtasks of each user,forming a joint optimization problem of task offloading and resource allocation.To solve the optimization problem,considering the bounded rationality and distributed decision-making mode of mobile users,an evolutionary game model is established.On this basis,in order to learn the strategies of other users and prevent the evolutionary stability strategy from falling into local optimization,a CORA＿EG algorithm combining Q-learning and differential evolution algorithm is proposed.Finally,the effectiveness of CORA＿EG algorithm is verified by experiments.Compared with the three benchmark methods,the computing cost can be saved by 72% to 85%.In addition,this method can seek the optimization strategy when other users’ decision information is incomplete,and it has robustness.