摘要：

面对信息化网络环境中大量时延敏感型和计算密集型任务的计算需求, 移动边缘计算(MEC)及其计算卸载技术提供了一种行之有效的解决方案。针对资源受限移动边缘系统的任务卸载策略, 设计一种成本最优化算法。首先, 结合系统的基本数据构建多用户多服务器网络场景, 并根据时延、能耗等待优化指标建立一种包含惩罚项的最小化成本优化模型; 然后, 提出一种改进人工蜂鸟算法(IAHA), 通过对原算法的寻优方式与算法结构进行适应性地调整和优化, 并引入一种紧急避险策略, 实现系统模型与算法映射的高度契合以及对模型问题快速精确求解, 进而得到系统的最优卸载策略; 最后, 应用策略进行部署以降低系统的成本支出和提升用户的服务体验。仿真实验结果表明, 所提改进算法能够有效降低系统成本, 并且在针对高维复杂模型求解时具有更突出的收敛性能和寻优精度, 在特定实验条件下, 所提改进算法相较于部分经典的元启发式算法和典型的新型群智能算法, 系统成本减少20.79%~65.39%, 所提任务卸载算法相对于本地计算策略的平均系统成本能够降低66.98%。

关键词: 移动边缘计算, 计算卸载, 卸载策略, 成本优化, 人工蜂鸟算法

Abstract:

Considering the computing requirements of a large number of delay-sensitive and computation-intensive tasks in the information network environment, Mobile Edge Computing (MEC) and its computation offloading technology provide an effective solution. Therefore, a cost optimization algorithm is designed for task offloading strategies in resource-constrained mobile edge systems. First, a multi-user and multi-server network scenario is constructed based on the basic data structure of the system, and a minimum cost optimization model, including penalty terms, is established based on optimization indicators such as latency and energy consumption. An Improved Artificial Hummingbird Algorithm (IAHA) is further proposed to adaptively adjust and optimize the structure and optimization method of the original algorithm, and an emergency avoidance strategy is introduced to achieve a high degree of fit between the system model and algorithm mapping, thereby providing a fast and accurate solution to the model problem and obtaining the optimal offloading strategy for the system. Finally, the application strategy is deployed to reduce system costs and enhance user service experience. The simulation results show that the proposed improved algorithm can effectively reduce system costs and has outstanding convergence performance and optimization accuracy when solving high-dimensional complex models. Under specific experimental conditions, this improved algorithm reduced system costs by 20.79% to 65.39%, respectively, compared with some classic metaheuristic and typical new swarm intelligence algorithms, and the average system cost is 66.98% less than those of local computing strategies with the proposed task offloading algorithm.

Key words: Mobile Edge Computing(MEC), computation offloading, offloading strategy, cost optimization, Artificial Hummingbird Algorithm(AHA)