面向用户公平性的边缘内容缓存策略  

Edge content caching strategy for user fairness in edge computing

作　　者：武继刚[1] 吴纯 陈龙[1] 吴亚兰[1] WU Jigang;WU Chun;CHEN Long;WU Yalan(School of Computer Science and Technology,Guangdong University of Technology,Guangzhou 510006,China)

机构地区：[1]广东工业大学计算机学院,广东广州510006

出　　处：《华中科技大学学报（自然科学版）》2022年第2期136-141,共6页Journal of Huazhong University of Science and Technology(Natural Science Edition)

基　　金：国家自然科学基金资助项目(62072118);广州市基础与应用基础研究项目(202102020248).

摘　　要：针对现有内容缓存研究中用户获取内容服务质量失衡导致的用户不公平问题,提出一种用户公平且快速高效的启发式内容缓存算法.在给出用户效用函数的基础上,首先,根据单位预算提升的最小效用及总效用定义内容和边缘服务器组合的优先级;然后,贪心地选择高优先级组合更新缓存策略.此外,以启发式算法的结果为初始解,定制一个模拟退火算法以进一步优化解的质量.实验结果表明:相比现有缓存方法,提出的两种算法可显著提升用户的最小效用,并能获得更高的公平指数.例如,针对多种存储空间配置的情况,与现有缓存方法相比,启发式算法和模拟退火算法分别可使用户的最小效用平均提升78.5%和87.3%,公平指数平均提高0.03和0.05.There exists unfairness of users for accessing to contents in the existing works on content caching,due to unbalance among users in terms of quality of service.Thus,an efficient heuristic algorithm was proposed to cache contents.Specifically,the utility function of users was defined.Then,the priority for combing contents and edge servers was also defined according to minimum utility and total utility for the increasement of a unit budget.The proposed heuristic algorithm updated caching strategy based on the priority of the combination of contents and edge servers.In addition,a simulated annealing algorithm was customized,to refine the solution generated by the heuristic algorithm.Experimental results show that,the proposed algorithms outperform the existing works on caching content,in terms of minimum utility of users and fairness among users.For example,for the cases of different storage capacities,the minimum utility of users for the two proposed algorithms can be improved by 78.5%and 87.3%on average,compared with the existing works.Meanwhile,the fairness indexes of users for two proposed algorithms increase 0.03 and 0.05 on average,compared with the existing works.

关 键 词：边缘计算 内容缓存 最大最小公平 启发式 模拟退火 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]