云边场景下基于合作博弈的数据上传优化  

Optimization of Data Upload Based on Cooperative Game Theory in Cloud Edge Scenarios

作　　者：钟传江 虞慧群[1,2] 范贵生 ZHONG Chuanjiang;YU Huiqun;FAN Guisheng(Department of Computer Science and Engineering,East China University of Science and Technology,Shanghai 200237,China;Shanghai Engineer Research Center of Smart Energy,Shanghai 201103,China)

机构地区：[1]华东理工大学计算机科学与工程系,上海200237 [2]上海智慧能源工程技术研究中心,上海201103

出　　处：《华东理工大学学报(自然科学版)》2025年第2期250-259,共10页Journal of East China University of Science and Technology

基　　金：国家自然科学基金(62372174,62276097)。

摘　　要：现有基于边缘计算的移动群智感知场景很少考虑边缘服务器之间的协作,而多边缘服务器协作面临边缘服务缓存、边缘设备资源约束和大规模边缘用户任务卸载的挑战。针对以上挑战,本文提出了一种面向多边缘服务器合作博弈和离散粒子群(MCG+DPSO)优化的计算卸载算法。该算法首先通过多边缘服务器之间的合作博弈(MCG)得到边缘用户任务在边缘服务器之间的中继初始方案。然后,将得到的初始方案作为离散粒子群算法(DPSO)的初始解。最后,通过DPSO算法得到最优解,实现用户任务和边缘服务器的匹配,从而最大化地降低移动感知平台数据上传的服务延迟和服务成本。通过在真实数据集上进行大量对比实验,结果表明与云策略、边缘间不通讯策略、随机策略、合作博弈策略、DPSO算法和差分进化算法相比,MCG+DPSO算法可以降低3.2%~56.0%的服务成本和3.6%~24.5%的服务延迟。In the existing mobile crowd intelligence sensing scenarios based on edge computing,the collaboration among edge servers is little considered.Collaboration among multiple edge servers faces challenges such as edge service caching,edge device resource constraints,and large-scale edge user task offloading.Addressing the above challenges,this paper formulates the optimization problem for service latency and service cost based on collaborative multi-edge server cooperation,and proposes a computational offloading algorithm based on multi edge server cooperative game and discrete particle swarm optimization(MCG+DPSO)optimization(MCG+DPSO).Firstly,an initial relay scheme for edge users’tasks among multiple edge servers is derived through Multi-Edge Server Cooperation Game(MCG).Then,the obtained scheme is taken as an initial solution for the Discrete Particle Swarm Optimization(DPSO)algorithm.Finally,the DPSO algorithm is used to obtain the optimal solution,achieving the matching between user tasks and edge servers,thereby maximizing the reduction of service latency and costs for data uploading on mobile sensing platforms.Through extensive comparative experiments on real datasets,it is shown that,compared with cloud strategy,edge non communication strategy,random strategy,cooperative game strategy,DPSO algorithm,and differential evolution algorithm,the proposed MCG+DPSO algorithm can reduce service costs by up to 3.2%to 56.0%and service latency by 3.6%to 24.5%.

关 键 词：边缘计算 合作博弈 移动群智感知 离散粒子群优化 任务卸载 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]