工程科学学报.第42卷.第7期：930-938.2020年7月 Chinese Journal of Engineering,Vol.42,No.7:930-938,July 2020 https://doi.org/10.13374/j.issn2095-9389.2019.07.12.001;http://cje.ustb.edu.cn 用户属性感知的移动社交网络边缘缓存机制 杨 静12,3，武佳2，)区，李红霞 1)重庆邮电大学通信与信息工程学院，重庆4000652)重庆高校市级光通信与网络重点实验室，重庆4000653)泛在感知与互联重庆市 重点实验室，重庆4000654)中国联合网络通信有限公司重庆市分公司，重庆401123 ☒通信作者，E-mail:1309431264@qq.com 摘要针对数据流量爆发式增长所引发的网络拥塞、用户体验质量恶化等问题，提出一种用户属性感知的边缘缓存机制. 首先，利用隐语义模型获知用户对各类内容的兴趣度，进而估计本地流行内容，然后微基站将预测的本地流行内容协作缓存， 并根据用户偏好的变化，将之实时更新.为进一步减少传输时延，根据用户偏好构建兴趣社区，在兴趣社区中基于用户的缓 存意愿和缓存能力，选择合适的缓存用户缓存目标内容并分享给普通用户.结果表明，所提机制性能优于随机缓存及最流行 内容缓存算法，在提高缓存命中率、降低传输时延的同时，增强了用户体验质量 关键词移动社交网络：边缘缓存：流行度预测：隐语义模型：社会属性 分类号TN929.5 User-aware edge-caching mechanism for mobile social network YANG Jing22,WU Jia,LI Hong-xid) 1)School of Communication and Information Engineering,Chongqing University of Posts and Telecommunications,Chongqing 400065,China 2)Chongqing Key Laboratory of Optical Communication and Networks,Chongqing 400065,China 3)Chongqing Key Laboratory of Ubiquitous Sensing and Networking.Chongqing 400065.China 4)Chongqing Branch of China Unicom,Chongqing 401123,China Corresponding author,E-mail:1309431264@qq.com ABSTRACT With the rapid growth in the number of intelligent terminal devices and wireless multimedia applications,mobile communication traffic has exploded.The latest report from Cisco Visual Networking Index(CVNI)indicates that by 2022,global mobile data traffic will have grown to three times that in 2017,which will exert tremendous pressure on the backhaul link.One key approach to solve this problem is to cache popular content at the edges(base stations and mobile devices)and then bring the requested content from the edges close to the user,instead of obtaining the requested content from the content server through backhaul networks.Thus,by obtaining the required content of mobile users locally,edge caching can effectively improve network performance and reduce the pressure on the backhaul link.However,owing to the limited storage capacity of the edge nodes and the diversification of user requirements,the edge nodes can neither cache all the content in the content server nor randomly cache the content.To solve these problems,an edge-caching mechanism based on user-awareness was proposed.First,using an implicit semantic model,we predicted popular content in a macro cell in terms of the users'interests.Small base stations within identical macro cells cache data cooperatively, which update local popular content based on the dynamic content preference of users.To further reduce the delay in content delivery,we helped users to ascertain their top communities of interest based on their content preferences.At the same time,the most appropriate user equipment(UE)is selected considering the caching willingness and caching ability to cache data for other UEs in identical communities of interest.Results show that the proposed mechanism outperforms the random cache approach and the most popular content-caching 收稿日期：2019-07-12 基金项目：国家自然科学基金资助项目(61771082.61871062：重庆市高校创新团队建设计划资助项目(CXTDX201601020)用户属性感知的移动社交网络边缘缓存机制 杨    静1,2,3)，武    佳1,2,3) 苣，李红霞4) 1) 重庆邮电大学通信与信息工程学院，重庆 400065    2) 重庆高校市级光通信与网络重点实验室，重庆 400065    3) 泛在感知与互联重庆市 重点实验室，重庆 400065    4) 中国联合网络通信有限公司重庆市分公司，重庆 401123 苣通信作者，E-mail: 1309431264@qq.com 摘    要    针对数据流量爆发式增长所引发的网络拥塞、用户体验质量恶化等问题，提出一种用户属性感知的边缘缓存机制. 首先，利用隐语义模型获知用户对各类内容的兴趣度，进而估计本地流行内容，然后微基站将预测的本地流行内容协作缓存， 并根据用户偏好的变化，将之实时更新. 为进一步减少传输时延，根据用户偏好构建兴趣社区，在兴趣社区中基于用户的缓 存意愿和缓存能力，选择合适的缓存用户缓存目标内容并分享给普通用户. 结果表明，所提机制性能优于随机缓存及最流行 内容缓存算法，在提高缓存命中率、降低传输时延的同时，增强了用户体验质量. 关键词    移动社交网络；边缘缓存；流行度预测；隐语义模型；社会属性 分类号    TN929.5 User-aware edge-caching mechanism for mobile social network YANG Jing1,2,3) ，WU Jia1,2,3) 苣 ，LI Hong-xia4) 1) School of Communication and Information Engineering, Chongqing University of Posts and Telecommunications, Chongqing 400065, China 2) Chongqing Key Laboratory of Optical Communication and Networks, Chongqing 400065, China 3) Chongqing Key Laboratory of Ubiquitous Sensing and Networking, Chongqing 400065, China 4) Chongqing Branch of China Unicom, Chongqing 401123, China 苣 Corresponding author, E-mail: 1309431264@qq.com ABSTRACT    With  the  rapid  growth  in  the  number  of  intelligent  terminal  devices  and  wireless  multimedia  applications,  mobile communication traffic has exploded. The latest report from Cisco Visual Networking Index (CVNI) indicates that by 2022, global mobile data traffic will have grown to three times that in 2017, which will exert tremendous pressure on the backhaul link. One key approach to solve this problem is to cache popular content at the edges (base stations and mobile devices) and then bring the requested content from the  edges  close  to  the  user,  instead  of  obtaining  the  requested  content  from  the  content  server  through  backhaul  networks.  Thus,  by obtaining  the  required  content  of  mobile  users  locally,  edge  caching  can  effectively  improve  network  performance  and  reduce  the pressure  on  the  backhaul  link.  However,  owing  to  the  limited  storage  capacity  of  the  edge  nodes  and  the  diversification  of  user requirements,  the  edge  nodes  can  neither  cache  all  the  content  in  the  content  server  nor  randomly  cache  the  content.  To  solve  these problems, an edge-caching mechanism based on user-awareness was proposed. First, using an implicit semantic model, we predicted popular content in a macro cell in terms of the users’ interests. Small base stations within identical macro cells cache data cooperatively, which update local popular content based on the dynamic content preference of users. To further reduce the delay in content delivery, we helped users to ascertain their top communities of interest based on their content preferences. At the same time, the most appropriate user equipment (UE) is selected considering the caching willingness and caching ability to cache data for other UEs in identical communities of interest. Results show that the proposed mechanism outperforms the random cache approach and the most popular content-caching 收稿日期: 2019−07−12 基金项目: 国家自然科学基金资助项目 (61771082，61871062)；重庆市高校创新团队建设计划资助项目 (CXTDX201601020) 工程科学学报，第 42 卷，第 7 期：930−938，2020 年 7 月 Chinese Journal of Engineering, Vol. 42, No. 7: 930−938, July 2020 https://doi.org/10.13374/j.issn2095-9389.2019.07.12.001; http://cje.ustb.edu.cn