密集异构蜂窝网中基于D2D的端到端多跳通信研究

Future ultra-dense heterogeneous cellular network, is able to achieve deep integration among aspects of frequency resources, temporal and spatial resources, and different communication modes, and to enable the cooperative multi-hop transmissions among D2D links and traditional cellular links; while bringing up new challenges to the areas of interference management, network topology, load balancing, inter-cell cooperation, QoS guarantee, etc. Furthermore, as the traditional Shannon information theory cannot apply, its impact on the actual per user achievable throughput capacity remains unknown. Towards this end, we aim to utilize available theoretical tools like stochastic geometry and queueing theory, to characterize the complicated dynamics experienced by data packets traveling from source to destination in the D2D communication based multi-hop routing; to develop various closed-form analytical models to derive the performance metrics of network throughput capacity, delay, end-to-end packet delivery probability, etc.; and to further study the optimum selection of D2D and eNB relays, the optimum frequency resource allocation, the optimum transmission mode selection, and the optimum transmit power control. The successful implementation of this project, is able to establish the fundamental basis for the design and optimization of future ultra-dense heterogeneous cellular network, and to provide both theoretical and technical support for the application of D2D communication technique in vehicular networks and Internet-of-things.

未来的密集异构蜂窝网络，在频谱资源、时空资源、通信模式等方面高度融合，将促使D2D链路与传统上下行cellular链路进行多跳协同传输；同时在干扰管理、网络拓扑、负载均衡、小区合作、用户QoS保障等领域都将提出全新的技术挑战。由于传统Shannon信息论无法适用，用户端到端的实际可达吞吐容量仍然处于未知。鉴于此，本研究拟结合随机几何等理论工具，建立数学模型刻画基于D2D通信的多跳传输中数据包从源端到目的端的随机传送过程；并显式求解吞吐容量、平均时延、端到端数据包传送概率等基本性能指标；在此基础上，进一步研究多跳传输路径中D2D和eNB中继的最优选择、多层eNB以及D2D通信之间频谱资源的最优分配、最优用户传输模式选择、最优传输功率控制等优化理论和方法。本项目的成功实施，可为未来密集异构蜂窝网的设计和优化打下必要的理论基础，并为D2D通信技术在车联网、物联网等的广泛应用提供理论和技术支持。

未来的密集异构蜂窝网络，在频谱资源、时空资源、通信模式等方面高度融合，将促使D2D链路与传统上下行cellular链路进行多跳协同传输；同时在干扰管理、网络拓扑、负载均衡、小区合作、用户QoS保障等领域都提出了全新的技术挑战。由于传统Shannon信息论无法适用，用户端到端的实际可达吞吐容量仍然处于未知。鉴于此，本研究结合机器学习、随机几何、凸优化、博弈论等理论和工具，对密集异构蜂窝网络中的D2D多跳通信，特别是基于COMP多点协作、D2D辅助NOMA通信、多层边缘服务器内容缓存及分发、多无人机空地协同通信等场景，建立数学模型刻画基于D2D通信的多跳传输中数据包从源端到目的端的随机传送过程，求解中断概率、覆盖率、平均可达速率等基本性能指标，并提出对应的网络控制参数优化方法。本项目的成功实施，可为未来密集异构蜂窝网的设计和优化打下必要的理论基础，并为D2D通信技术在车联网、物联网等的广泛应用提供理论和技术支持。