基于内容拓扑的网络化协作缓存技术研究

Along with recent advances in mobile communication technologies, an ever growing number of mobile users are continuously attracted to enjoy a wide plethora of multimedia services using smartphones and tablets. Accordingly, the demand for rich multimedia services over mobile networks has been soaring at a tremendous pace over recent years, which has been regarded as one of the main challenges facing the future 5G wireless communications. In this project, to cater for this trend, we aim at fully exploiting the benefit promised by jointly utilizing the cache and communication resource in cellular networks. Particularly, we will, from the multi-time scale perspective, propose efficient caching strategies for multicell-coordination enabled cellular networks, where neighbor small cell BSs (SBSs) are allowed to collaborate for content share with high-speed links. Firstly, regarding content as the new resource in wireless networks, we study the complicated coupling relationship among the physical connection topology, interaction among cooperative links, content distribution and cooperative transmission costs. In this light, both the novel definition and construction method of "content topology" are proposed and then, the content placement (proactive caching) strategy is developed, with which the content based graph connectivity (CGC) can be guaranteed. Based on this large-time scale management of content distribution, we further study the propagation feature of the newly-coming content and reveal its essential effect on the popularity of the whole content files. Moreover, we have developed an on-line content update strategy, with which the CGC can be guaranteed and meanwhile the update cost can be minimized. These achievements of this project will provide advanced theoretical and key technical supports for providing users with better experiences of multimedia services in future 5G wireless networks.

随着无线传输技术的不断发展以及智能移动终端设备的日益普及，全球移动业务数据呈现出爆发式增长，对未来5G无线网络的承载能力提出了巨大挑战。本项目旨在挖掘异构小蜂窝网络中存储资源与传输资源的联合调度增益，从网络化协作的角度研究内容分布式缓存机制，实现局部存储资源的区域化、网络化共享，应对未来海量移动数据的承载需求：首先，探究小蜂窝基站之间物理拓扑结构、协作链路间互扰关系、内容存储位置以及协作传输开销之间的复杂耦合关系，提出网络“内容拓扑”的定义及构造方法，进而设计保障内容连通度的内容放置策略，完成内容分布的大时间尺度“管理”。在此基础上，探究突发内容在网络中的传播特征，揭示其对于整体内容流行度时空变化的作用规律，在保障网络内容连通度的同时设计最小开销的内容更新策略，完成内容分布的小时间尺度“优化”。本项目的研究可为5G无线网络中多媒体内容的高效承载提供理论与技术支撑。

随着无线传输技术的不断发展、智能移动终端设备的日益普及以及人们对实时应用需求的不断增强，全球移动业务数据呈现出爆发式增长态势，对异构无线网络设备以及网络整体的数据承载能力提出了巨大挑战。本项目着眼于挖掘网络中存储资源与传输资源的联合调度增益，实现局部存储资源的区域化、网络化共享，从而在有效缓解数据流量压力的同时保障数据消费者所获得数据信息的时效性。首先，课题组针对异构蜂窝网络中多媒体内容的网络化缓存问题，通过综合考虑用户喜好、内容存储位置、传输功率以及链路之间干扰情况等因素，提出了内容拓扑的概念以及构造方法，并以此为基础提出了基于内容放置、传输链路激活以及传输功率控制的分布式联合优化策略，有效提升了基站处的流量卸载比率，改善了资源利用效率。进而，以网络中实时应用对于数据时效性的需求为出发点，一方面，揭示了内容流行度、内容放置策略以及复杂网络参数对于数据新鲜度损失、传感器能耗的影响规律，设计了一种适用于大规模网络的内容放置策略，能够在满足用户对数据新鲜度损失需求的同时，有效降低传感器的能量消耗，实现差异化内容的大时间尺度“管理”；另一方面，进一步针对实际网络中用户需求存在动态变化且无法预知的特性，探究了数据消费者以及缓存节点处信息年龄的动态演变过程，设计了基于深度强化学习的内容动态更新策略，有效提升了内容动态更新过程所带来的平均收益，实现了内容分布的小时间尺度“优化”。基于本项目相关研究成果，课题组发表学术论文7篇，其中主持人以第一作者/通信作者身份发表学术论文6篇；此外，基于相关成果，授权发明专利2项。本项目的研究可为未来无线网络中数据内容的高效缓存与共享提供理论与技术支撑。