未来5G流媒体服务中知识驱动的缓存与交付机制研究

In the Future, the mobile video streaming service in 5G wireless network with ultra-low latency, high quality of experience, large scale and personality heavily depends on effective ubiquitous caching and data delivery. However, current solutions are suffering from the unawareness of complex network environment, dynamic user behavior and diversified service characteristics. These drawbacks not only lead to the underutilized of edge caching resources but also make it difficult to ensure the video delivery efficiency and quality. However, current solutions are suffering from the unawareness of complex network environment, dynamic user behavior and diversified service characteristics. These drawbacks not only lead to the underutilized of edge caching resources but also make it difficult to ensure the video delivery efficiency and quality. As a result, the demand of video streaming service in 5G scenarios cannot be fulfilled. To solve these problems, this project aims to study the knowledge-driven content caching and delivery for 5G video streaming: 1. Establishing a multi-dimensional 5G net streaming media modeling. Constructing knowledge description system for networks, clients and services; 2. Researching on knowledge definition joint optimization of streaming media cache and delivery. Solving the synergy problem of cache and delivery, and achieving the deep integration of cache and delivery; 3. Designing a high efficient ubiquitous cache for mass and personalized content and optimizing the whole network cache distribution in real time; 4. Proposing smart video delivery over 5G ultra density heterogeneous network to improve forwarding efficiency and ensure delivery quality. Based on above researches, knowledge-based streaming media cache and delivery over 5G network can be realized, the content acquisition delay can be significantly reduced, the network carrying capacity can be improved, and the reliable personalized streaming service can be supported.

未来5G场景下超低时延、高质量、大规模、个性化的流媒体服务依赖于高效的内容缓存与交付。然而传统解决方案由于无法立体认知复杂网络环境、用户行为和业务特性，且缓存与交付缺乏协同，造成网络边缘节点的缓存资源难以充分利用，流媒体交付效率以及质量得不到有效保障，难以适应5G网络下流媒体服务发展需求。为此，本项目拟开展未来5G流媒体服务中知识驱动的缓存与交付机制研究。具体包括：1.建立多维立体的5G网络及流媒体知识模型，构造面向网络、用户及业务的知识描述体系；2.研究知识定义的流媒体缓存与交付联合优化，解决缓存与交付协同问题，实现缓存与交付深度融合；3.设计面向海量、个性化内容的高效泛在缓存，实时优化全网缓存分布；4.提出5G超密异构网络的流媒体智慧交付机理，提高转发效率并保障交付质量。通过上述研究，实现5G网络下知识化的流媒体缓存与交付，大幅降低内容获取时延、提高网络承载能力并保障高质量流媒体服务。

随着移动通信技术的发展，流媒体服务已深刻融入人们的生活，并成为了解讯息、寻求共鸣的重要应用。然而，5G网络边端设备异构、服务过程复杂；同时，网络资源呈现碎片化分布且利用率不足。在面对下一代超高清视频、沉浸式媒体等新兴媒体形式时，现有的分发技术仍然捉襟见肘，严重阻碍了流媒体服务质量的进一步提升。为此，本项目从未来5G流媒体的缓存与交付入手，开展前瞻性研究，主要研究内容如下：（1）设计多维立体的5G网络及流媒体知识模型，包括多尺度5G流媒体用户知识表征方法、异构网络演化知识提取及建模方法。（2）知识定义的流媒体内容缓存与交付联合优化框架，包括知识模型与5G流媒体服务深度关联机制、流媒体内容缓存与交付的协作适配机理等。 （3）面向海量、个性化流媒体内容的高效泛在缓存方法，包括面向流媒体内容感知的实时编码缓存策略、分布式的主/被动5G网络边缘节点缓存机制等。（4）5G超密异构网络下的流媒体智慧交付机理，包括以知识为中心的5G网络内容转发平面构建机制、动态自适应的移动视频交付质量保障方法等。上述研究工作为未来5G流媒体服务提供了一整套行之有效的解决方案。经过4年的研究，项目组共计发表高质量论文23篇，其中SCI检索期刊论文17篇，国际知名会议论文6篇，超额完成原计划发表12篇学术论文的任务。其中部分论文发表在多媒体通信领域权威期刊IEEE Journal on Selected Areas in Communications,IEEE Internet of Things Journal,IEEE Transactions on Mobile Computing以及若干IEEE Trans.级别期刊，得到国际众多知名学者的关注和积极评价。申请专利8项都已授权。项目在研期间，项目组成员多次参加国际学术交流会议，邀请国外专家开展线上学术研讨，促进了国际间的学术交流。项目负责人担任了国际权威SCI期刊Transactions on Emerging Telecommunications Technologies (Wiley)主编，获得了2022年国家自然科学基金杰出青年基金资助，项目执行期间培养了博士毕业生7名，硕士毕业生11名，超额完成任务。