基于物理-社交信息的D2D内容共享技术研究

Massive content sharing has become the main challenging problem for the next-generation cellular networks, especially for the joint impact of the dynamic communication environment, heterogeneous smartphones, and various application services. As such, this project investigates the D2D content sharing technology based on the joint physical and social information, by exploring the potential of mobile devices and improving the performance of massive content sharing. Specifically, this project analyzes the physical layer transmission performance and the social characteristic performance in the framework of cache-assisted D2D communication framework, in terms of the throughput and the social tie strength, respectively. Next, their integrated relationship is explored to keep the additive advantage. In this way, from the aspect of solving the massive content preset, matching and transmission problems, this project proposes distributed cache management scheme, including the joint scheme for distributed cache decision and update and the coordinated cache scheduling, so as to improve the cache efficiency. For the purpose of improving election efficiency and obtaining a win-win solution, this project proposes the distributed content distribution scheme based on the physical and social domains and matching game. In order to deal with the uncertainties and avoid the interference problem, this project proposes the robust resource management optimization mechanism based on the joint evolution game and differential game. In this way, this project is potential to provide a powerful solution for realizing efficient D2D content sharing.

针对下一代蜂窝网络中海量内容共享难题，尤其是存在的诸如通信环境复杂多变、用户终端互异、内容服务需求多样化等多重异质条件，本项目以充分挖掘移动终端侧的潜力、改善海量内容共享性能为目标，开展基于物理-社交信息的D2D内容共享技术研究，具体包括：分析缓存辅助的D2D通信框架下以吞吐量为代表的物理层传输性能和以社交关系强度为代表的社交属性性能，并从确保两者的叠加优势出发，探寻两者的融合关系；在此指导下，从解决海量内容的预置、匹配和传输问题入手，以提升缓存效率为目的，提出包括分布式缓存决策与更新的联合优化、缓存内容的协同调度在内的动态分布式缓存管理方案；以提高互选效率，确保共赢为目的，提出基于物理-社交域和匹配博弈的分布式内容分发方案；以有效应对网络环境的多重不确定因素，避免棘手的干扰问题为目的，提出联合演化博弈和微分博弈的鲁棒资源管理优化机制，从而为实现高效的D2D内容共享提供有力的解决途径。

本项目考虑到下一代蜂窝网络中存在的诸如需求多样化等多重异质条件，以及移动终端缓存能力和传输能力受限的现实，构建了缓存辅助D2D通信框架。在此框架下，从内容预置、匹配和传输等不同层面为突破口，凝练出物理信息度量和社交信息度量的融合问题、多重异质和不确定条件下的内容预存与适配问题，以及复杂动态环境下高效内容传输的鲁棒性问题三个关键科学问题。以解决这三大问题为突破口，所研究的基于物理-社交信息的D2D内容共享技术为解决下一代蜂窝网络中的海量数据传输难题提供了崭新的研究思路。主要解决的关键技术与创新点总结如下：1）提出了基于随机几何的物理层传输性能分析方法和基于社交关系强度学习模型的社交关系度量表征方法，探究了两者间的融合机理，为面向D2D通信的内容共享技术提供了理论指导。2）挖掘了社交关系强度在增强缓存管理本地特性上的优势作用，提出了多重异质条件下缓存自适应管理与协同调度方法，从而在移动终端缓存容量互异且受限的条件下，有效提升了缓存效率，确保了内容共享的并发能力和可扩展性。3）构建了基于离散时间马尔科夫链的用户移动性模型和社交中心度推断模型，提出了物理-社交域融合下基于三维匹配的内容分发方案，在内容请求者和提供者之间实现了具有较高互选效率和较低复杂度的互配过程。4）明析了社交关系强度的稳定性和合作激励在资源管理优化问题上的作用机理，设计了基于随机学习和博弈理论的面向D2D组播内容推荐的资源管理优化方法，确保了多重不确定因素下共享内容的正确传输。