面向时延和能耗联合优化的MEC计算卸载策略

Joint optimization of delay and energy consumption computation offloading scheme for MEC

在移动边缘计算(MEC)中,计算卸载可以有效缓解资源受限和提高网络服务质量。以任务执行时延、终端能耗和边缘服务器负载率的联合优化为目标,提出面向时延和能耗联合优化的MEC计算卸载策略。构建多目标约束的成本优化模型,引入多变异算子,以迭代关联概率更新变异算子,设计多变异差分进化(MDE)算法求解,实现计算卸载成本最优。为验证MDE算法的有效性,基于Autonomous Systems by Skitter公开数据集构建3个不同规模的实验网络,将MDE算法与随机计算卸载算法、能量优化计算卸载算法、多目标贪婪计算卸载等算法进行对比分析,MDE算法的执行成功率、卸载成功率、服务器负载均衡性分别平均提升了13.23%,12.96%,29.37%,MDE算法能实现MEC中高效、稳定的计算卸载。

In Mobile Edge Computing(MEC),computation offloading can effectively alleviate resource constraints and improve network quality of service.A joint optimization of delay and energy consumption computation offloading scheme for MEC was proposed with the joint optimization of task delay,terminal energy consumption and edge server load rate.A cost optimization model with multi-objective constraints was constructed,and a Multi-Mutation Differential Evolution(MDE)algorithm was designed by introducing the multi-mutation operator that was updated with iterative correlation probability to solve.To verify the effectiveness of MDE algorithm,three different scale experimental networks were constructed based on Autonomous Systems by Skitter public dataset.Compared with random computation offloading scheme,energy optimization computation offloading scheme and multi-objective greedy computation offloading scheme,MDE algorithm improved the execution success rate,offloading success rate and server load balancing by 13.23%,12.96%,29.37%respectively,which could realize efficient and stable computation offloading in MEC.