超蜂窝网络基于用户行为预测的软实时服务机制与能效优化

发布时间：2018-06-25 16:48

本文选题：内容服务 + 流行度预测　；参考：《中国科学:信息科学》2017年05期

【摘要】：无线蜂窝网络中相同的数据内容被大量重复传输,造成了网络能耗的增加和无线资源的浪费.然而,现有无线蜂窝网络并不关注所传输数据的实质内容,并对所有移动用户提供无差别服务,因此无法解决这一传输冗余问题.在分析移动用户行为的基础上,结合无线蜂窝网络和数字广播系统的优势,提出了以用户为中心、具有数据业务内容动态感知能力的软实时服务机制,降低无线蜂窝网络的传输能耗和无线资源消耗,同时提升移动用户体验的服务质量.本文给出了业务内容流行度预测、用户信息访问行为预测、动态多播小区构建与内容推送等算法设计,并建立了系统的性能模型,揭示了系统参数与系统性能之间的关联.经过公共网络新闻类信息访问的实测数据模拟验证,在理想的预测算法下,广播排名前10%的热门网络内容,可将蜂窝网络中的数据传输能效提升约2倍;经过校园网络在线学习场景的两个数据集模拟验证,在理想预测算法下,可将网络数据传输能效提高50倍.
[Abstract]:The same data content in wireless cellular networks is transmitted repeatedly, which results in the increase of network energy consumption and the waste of wireless resources. However, the existing wireless cellular networks do not pay attention to the substance of the transmitted data, and provide non-differentiated services to all mobile users. Therefore, this problem of transmission redundancy cannot be solved. Based on the analysis of mobile user behavior and the advantages of wireless cellular network and digital broadcasting system, a new soft real-time service mechanism, which is user-centered and has the capability of dynamic sensing of data service content, is proposed. The transmission energy consumption and wireless resource consumption of wireless cellular network are reduced, and the quality of service of mobile user experience is improved. In this paper, the algorithm design of business content popularity prediction, user information access behavior prediction, dynamic multicast cell construction and content push are presented. The performance model of the system is established, and the relationship between system parameters and system performance is revealed. Through the simulation and verification of the measured data accessed by the public network news information, under the ideal prediction algorithm, broadcasting the top 10% of the hot network content can increase the efficiency of data transmission in the cellular network by about 2 times. Through the simulation of two data sets in the online learning scene of campus network, the efficiency of network data transmission can be improved by 50 times under the ideal prediction algorithm.
【作者单位】：清华大学电子工程系;清华大学信息科学与技术国家实验室(筹);浙江大学电信工程系;
【基金】：国家重点基础研究发展计划(973)(批准号:2012CB316000) 泛网无线通信教育部重点实验室开放课题(批准号:KFKT-2014101)资助项目
【分类号】：TN929.5

【相似文献】