基于多群体混合智能优化算法的卸载决策寻优方法  

Unloading decision optimization method based on multi-population hybrid intelligent optimization algorithm

作　　者：方浩添 田乐[1] 郭茂祖 FANG Haotian;TIAN Le;GUO Maozu(Beijing Key Laboratory of Intelligent Processing of Building Big Data,Beijing University of Civil Engineering and Architecture,Beijing 102600,China)

机构地区：[1]北京建筑大学建筑大数据智能处理方法研究北京市重点实验室,北京102600

出　　处：《智能系统学报》2024年第6期1573-1583,共11页CAAI Transactions on Intelligent Systems

基　　金：国家自然科学基金项目(62271036);国家重点研发计划科技冬奥重点专项(2021YFF0306303).

摘　　要：在移动边缘计算的网络架构中,为权衡降低计算应用卸载的能耗与时延,引入卸载决策控制器,并通过卸载决策寻优算法得到最优卸载决策。结合人工蜂群算法和人工鱼群算法提出新的人工蜂-鱼群(artificial bee colony-fish swarm,ABC-FS)算法,在此基础上引入高斯衰减函数将算法参数由静态变为动态,并将改进粒子群算法的惯性权重因子引入算法中,从而得到一种多群体混合智能优化算法;设计联合优化时延与能耗的目标函数,再依据泊松概率进行仿真实验。仿真实验结果表明,提出的卸载策略寻优算法,与多组对照组相比,收敛速度更快,且在多接入边缘计算的场景下能权衡降低系统中任务卸载的总时延与总能耗。In the network architecture of mobile edge computing,an offloading decision controller was introduced to balance the reduction of energy consumption and delay.This controller obtains the optimal offloading decision through an offloading decision optimization algorithm.A new ABC–FS algorithm was proposed by combining the artificial bee colony(ABC)algorithm and the artificial fish swarm(FS)algorithm.Additionally,a Gaussian decay function was introduced to transition the algorithm parameters from static to dynamic,and the inertia weight factor of the improved particle swarm optimization algorithm was incorporated,creating a multi-population hybrid intelligent optimization algorithm.Finally,an objective function that jointly optimizes delay and energy consumption was designed,and simulation experiments were conducted using Poisson probability.Simulation results show that the proposed offloading strategy optimization algorithm achieves faster convergence speed compared to several benchmark methods and effectively balances the reduction of total task offloading delay and total energy consumption in multi-access edge computing scenarios.

关 键 词：移动边缘计算 计算卸载 人工鱼群算法 人工蜂群算法 自相似排队模型 高斯衰减函数 粒子群算法 惯性权重因子 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]