改进多种群进化算法求解移动边缘计算中任务调度问题

移动边缘计算通过在靠近用户端的网络边缘部署服务器，为用户提供低时延的网络通信服务和类似云的计算服务。移动设备通过网络接入点将任务卸载到边缘服务器进行处理，能够有效地减少移动设备的能耗以及任务的完成时间。然而，用户在卸载任务时需要支付一定的通信成本。本文在构建包含多个用户和多个边缘计算节点的移动边缘计算环境的基础上，建立了最小化移动设备的任务完成时间、能耗以及通信成本的数学模型。为了解决上述问题，本文提出了一种改进多种群进化算法的任务调度优化算法。该调度算法通过优化卸载决策和资源分配决策来达到降低移动设备综合成本的目的。大量仿真实验说明，该任务调度算法与其他几种的任务调度算法相比，能够更有效地降低移动设备的综合成本。

An Improved Multi-population Evolutionary Algorithm for Task Scheduling in a Mobile Edge Computing Environment

Mobile edge computing (MEC) can provide users with low-latency network services and cloud-like computing services by deploying servers at the edge network which is close to users. Mobile devices (MDs) offload their tasks to edge servers for computing via the network access points, which can effectively reduce the power consumption of MDs and the completion time of their tasks. However, users have to pay for communications when they offload their tasks to edge servers. A MEC system is studied which contains multiple users and multiple edge computing nodes. Mathematical models are built for task completion time, power consumption, and communication cost of MDs, and the problem is formulated to minimize these objectives. A task scheduling algorithm based on a multi-population evolutionary algorithm is proposed to solve this problem. The scheduling algorithm minimizes the comprehensive cost of MDs by optimizing the offloading decisions and resource allocation decisions for MDs. Lots of simulations are conducted to verify that the proposed algorithm can reduce the comprehensive cost of MDs better compared with other scheduling algorithms.