不完全网络状态信息下D2D通信干扰协调及实时资源调度研究

Device-to-Device communication is a novel communication technology in the fifth-generation mobile system. It can efficiently relieve the traffic burden of the base station, improve the system spectral efficiency and throughput. To handle the difficulty of obtaining the complete network state information of D2D communications, investigate the interference coordination and real-time resource scheduling problems of D2D communications with incomplete network state information. First, propose a method combining the binary state feedback and belief function to predict the link state, which will reduce the communication overhead of obtaining the system state information. Then, using the swarm intelligent method to investigate the interference coordination problem under shared resources, by flexibly assigning channels and powers with supporting of multiple D2D users sharing the same resource, to develop the potential of D2D communications in improving the system throughput. Using the multi-armed bandit model to investigate the real-time resource scheduling problem under dedicated resources, to handle the challenges caused by the uncertain factors and heterogeneous deadline requirements. By investigating the indexability of the problem and deriving the closed-form expression of the index, the research can further propose a high index first scheduling algorithm with low complexity. Finally, study the spectrum partition problem of shared and dedicated resources based on the sequential decision process, to optimize the spectral efficiency and guarantee the system throughput and real-time communication requirements. The researches will contribute on improving the throughput of the D2D communication system and its ability to transmit the real-time traffic.

D2D通信是5G中一种新型的通信方式，可有效缓解基站流量负担、提升系统频谱效率和吞吐量。针对D2D通信中完全网络状态信息获取难的问题，研究不完全网络状态信息下D2D通信干扰协调及实时资源调度问题。首先，提出二进制状态反馈和信度函数相结合的新型链路状态表征方法，降低获取链路状态信息的通信开销。然后，借鉴群体智能思想研究共享资源下的干扰协调问题，通过灵活的信道和功率分配以支持多D2D用户复用资源，挖掘D2D通信在提升系统吞吐量方面的潜力；利用多臂赌博机模型研究专用资源下的实时D2D通信资源调度问题，应对不确定性因素和异构实时需求带来的挑战，通过探究问题可索引特征和索引值解析表达式，提出高索引值优先的低复杂度调度算法。最后，基于序贯决策过程研究共享资源与专用资源的频谱划分问题，优化频谱效率并兼顾系统吞吐量和用户实时传输需求。本课题研究有助于提升D2D通信系统吞吐量和其承载实时业务的能力。

D2D通信支持直连通信，信息无需经过基站直接在临近用户之间交互，可有效缓解基站的流量负载。通过有效的干扰协调方法和资源调度方法可提升系统吞吐量和频谱利用率。由于获取D2D通信网络完全网络状态信息开销大，本项目针对不完全网络状态信息下D2D通信干扰协调及实时资源调度问题开展了研究。首先，研究了信息不完全网络状态的表征方法和系统吞吐量的评估方法，并提出了蜂窝用户和D2D用户功率及信道分配算法，有效协调了用户之间的干扰，提升了系统吞吐量；其次，研究了实时D2D通信资源管理方法及不完全网络状态场景下的D2D用户调度方法，有效保证了D2D用户的实时通信需求和吞吐量需求；最后，探究了网络系统中同时存在吞吐量和实时性要求场景下的资源管理方法，通过严密的理论分析及推导给出了稳定最优及随机最优资源管理方法，可有效满足用户服务质量要求。本项目取得的研究成果基于严密的理论推导，并有相应的实验验证，所提出的方法可有效协调用户之间干扰，应对不完全网络状态信息带来的挑战，提升用户吞吐量需求并保证实时传输需求。