Task Offloading And Resource Allocation With Deadline Constraints In Heterogeneous Edge Computing Environment

Edge computing is widely used in emerging fields such as Internet of Things,Internet of Vehicles,and online games.Edge computing provides low-latency computing services for terminal devices by deploying computing resources at the edge of the network.Because single edge node with limited computing resources cannot execute large-scale tasks,in order to minimize the total tardiness of edge computing system,multi-edge coordinated scheduling needs to be considered.However,optimization problems such as task offloading and resource allocation with deadline constraints in a heterogeneous edge computing environment have the following challenges:(i)In a heterogeneous edge computing environment,edge nodes have different computing and transmission capabilities.How to achieve a balance between task execution time and transmission time when making task offloading decisions is a challenge problem.(ii)Each edge node with limited resources needs to service multiple tasks.Tasks have different deadline constraints.How to minimize the total tardiness of edge computing system under the premise of meeting the task deadline constraints is much difficult.Considering the problem of task offloading in the multi-edge collaborative environment,a multi-edge collaborative task offloading method is proposed.The task offloading method uses deadline approaching rule to prioritize the tasks waiting to be offloaded.After completing the prioritization,a task offloading decision is made for each task,and the impact of task execution time and transmission time on the task offloading decision is comprehensively considered.With the goal of minimizing the total tardiness of edge computing system,the task offloading decision is made based on the greedy selection method.The edge node with the earliest estimated completion time is selected as the execution location of the task.After the task is transmitted to the destination edge node,the earliest completion time priority rule is called to select a specific virtual machine from the current edge node to execute the task.For the problem of task offloading and resource allocation with deadline constraints in a heterogeneous edge computing environment,a heterogeneous edge collaborative task offloading and resource allocation method is proposed.It contains five components: edge network topology node sequencing,offloading task sequencing,task offloading decision sequence generation,resource allocation and scheduling result adjustment.Considering the execution time and transmission time of the task comprehensively,three offloading decision methods are proposed that based on the earliest estimated completion time,the maximum offloading revenue and the most recent minimum load respectively.The method introduces three task sorting rules: first come first served,deadline approaching priority,and minimum slack time priority.For the problem of multi-task competition,allocates resources reasonably,three virtual machine selection rules are put forward: the earliest completion time priority,the earliest available time priority,and load balancing priority.In order to solve the impact of uneven task distribution on scheduling in a heterogeneous edge environment,considering the computing power and load pressure of each edge node,three rules based on node load,node degree and node idleness are proposed to determine the processing order respectively.The method adopts computational intensity and estimated slack rules to get the order of task offloading.The initial scheduling is optimized by the improved particle swarm algorithm.And the total tardiness of edge computing system is minimized.To evaluate the performance of the proposed algorithm,several comparison algorithms are selected,and the performance differences between algorithms are analyzed with the help of multi-factor analysis of variance technique.For the task offloading problem of multi-edge collaboration,the results of the proposed algorithm are significantly better than other comparison algorithms under different number of edge nodes,task numbers and deadlines.When resources are sufficient,the proposed algorithm has a outstanding performance.For the task offloading and resource allocation problems with deadline constraints in a heterogeneous edge computing environment,the parameters of algorithm are calibrated first and the optimal parameter combination is selected.Three comparison algorithms are selected for comparison.The performance differences between algorithms are analyzed from the perspectives of different number of edge nodes,number of tasks,distribution of tasks,and interval of deadlines.The experimental results show that the proposed algorithm outperforms other comparison algorithms.The algorithm that considers task offloading and resource allocation comprehensively is more effective than the algorithm that only makes task offloading decisions when the resources are sufficient and the deadlines of the tasks are not strict.