流行度感知的无线视频云边缓存策略研究  

Popularity-aware cloud-edge collaborative caching strategy for wireless video

作　　者：唐汉秦 赵辉[1,2,3] 宁竞莜 王静 万波[1,3] 王泉 TANG Hanqin;ZHAO Hui;NING Jingyou;WANG Jing;WAN Bo;WANG Quan(School of Computer Science and Technology,Xidian University,Xi’an 710071,China;Hangzhou Institute of Technology,Xidian University,Hangzhou 311231,China;Key Laboratory of Smart Human-Computer Interaction and Wearable Technology of Shaanxi Province,Xi’an 710071,China)

机构地区：[1]西安电子科技大学计算机科学与技术学院,陕西西安710071 [2]西安电子科技大学杭州研究院,浙江杭州311231 [3]陕西省智能人机交互与可穿戴技术重点实验室,陕西西安710071

出　　处：《西安电子科技大学学报》2024年第5期97-109,共13页Journal of Xidian University

基　　金：陕西省重点研发计划(2024GX-YBXM-010,2024GX-YBXM-140,2024GX-YBXM-039);陕西省创新团队(2023-CX-TD-08);陕西省秦创原“科学家+工程师”队伍(2023KXJ-040);中央高校基本科研业务费专项资金(ZYTS24089)。

摘　　要：移动边缘缓存技术将视频缓存在离用户更近的边缘服务器,从而为用户提供更加便捷的服务。目前的视频缓存方法主要基于整体的视频流行度,忽视了视频流行度在时空上的差异,未能充分利用边缘服务器的广地域分布特性,影响云边环境下视频缓存的效果。针对此问题,笔者提出了基于流行度感知的无线视频云边缓存策略。首先,基于分布式协作的云边架构,考虑视频流行度在时空上的差异性,并结合视频分片及视频片段流行度,以最小化所有用户请求视频的平均时延和最大化用户请求视频的缓存总命中率为目标,建立云边视频缓存模型。其次,针对边缘服务器有限的计算资源和缓存资源,提出一种基于全局价值评估的缓存策略,将某一视频片段满足用户请求的能力表示为缓存价值,同时引入缓存价值惩罚机制,动态完成缓存内容的价值评估,实现视频片段高效缓存。最后,通过仿真实验证明所提出的策略能够显著地降低平均传输时延和回程流量负载,提高缓存资源的命中率。Mobile edge caching technology caches videos at edge servers closer to users,thereby providing users with more convenient services.Current video caching methods rely primarily on overall video popularity,ignoring the spatial and temporal variations in video popularity.As a result,they fail to fully utilize the wide geographical distribution characteristics of edge servers,thereby impacting the effectiveness of video caching in the cloud-edge environment.In order to address this issue,we propose a wireless video cloud edge caching strategy based on popularity perception.First,based on the cloud-edge collaborative architecture,we establish a cloud-edge video caching model that takes into account the spatial and temporal variations in video popularity.This model combines video segmentation and video segment popularity,and aims to minimize the average delay for all videos requested and maximize the total cache hit rate.Second,considering the limited computational and caching resources of edge servers,we propose a caching strategy called Global Value Evaluation(GVE).This strategy quantifies the ability of a video segment to fulfill user requests as its caching value and incorporates a caching value penalty mechanism to dynamically assess the value of cached content,enabling efficient caching of video segments.Finally,simulation experiments demonstrate that the proposed strategy can significantly reduce the average transmission delay and backhaul traffic load,and improve the cache hit rate of requested videos.

关 键 词：移动边缘计算 视频点播 视频缓存 云边协作 视频流行度 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]