面向海量连接的5G蜂窝网网间频谱共享方法研究

The Internet of Things (IoT) based on cellular networks has become an important strategic initiative for all countries in the world. However, with the influx of massive devices, the scarcity of the spectrum resource in cellular networks will turn to be the main bottleneck restricting its development. Under the current static spectrum management policy, a large amount of spectrum is in idle state. Therefore, this project will investigate the inter-network spectrum sharing scheme with the consideration of the characteristics of IoT, such as the massive connectivity, the low cost and low power feature of devices, the diversity of services, etc., to bring the idle spectrum resource of other dedicated networks into future cellular networks for the data traffic delivery. To be specific, first, from the network level, we will design a cognitive mesh cellular network architecture, which can apply to the inter-network spectrum sharing, to solve the problem that the shared spectrum bands are not compatible with IoT devices and the spectrum saturation problem caused by the massive direct connections to each base station. Second, we will analyze the spectrum demand of an operator with certain specific communication scenario by considering the uncertainty of both data traffic and spectrum availability, and develop a multi-unit combinatorial auction mechanism and a Stackelberg game based pricing mechanism to allocate the shared spectrum among different operators effectively. Finally, we will design data transmission schemes by considering different service types and different spectrum features for each operator to achieve an efficient utilization of the spectrum. The above works will effectively solve the bottleneck problem of spectrum shortage, which can promote the development of the IoT, and also provide theoretical supports for both innovation on spectrum management strategy and improvement on spectrum utilization.

以蜂窝网为载体的物联网建设已成为各国的重要战略举措，然而，目前蜂窝网的频谱资源已严重匮乏，随着未来海量设备涌入，频谱短缺将成为制约其发展的主要瓶颈。考虑到当前静态频谱管理策略导致的频谱闲置现象，本课题将综合物联网海量连接、设备低成本低功耗、业务多样等特性，针对网间频谱共享展开研究，将其他各类专网（外网）的空闲频谱用于未来蜂窝网的数据传输：1设计适用于网间频谱共享的网状认知蜂窝网架构，从网络层面解决设备不支持外网频段问题及海量设备直连基站导致的频谱饱和问题；2考虑流量和共享频谱的不确定性，基于实际通信场景分析运营商的频谱需求，提出多商品捆绑式拍卖机制和基于斯塔克尔伯格博弈的定价机制，通过经济模型实现对共享频谱的合理分配；3基于业务类型和频谱属性设计频谱节约式传输方案，实现对共享频谱的合理高效利用。本课题将有效克服频谱瓶颈制约，促进物联网发展，为创新频谱管理模式、提高频谱利用率提供理论支持。

万物互联时代的到来使得频谱的供需矛盾日益突出，本项目针对频谱短缺对下一代移动通信网络发展的掣肘，探索基于动态频谱共享的解决方案，从网络使能架构、频谱使用权分配、共享频谱规划等方面展开一系列深入研究，取得如下重要成果：1、提出网状认知蜂窝网络架构和授权式频谱共享机制，并设计了前瞻式和后验式两种频谱贸易方案，通过经济激励手段促成主、次服务商之间的动态频谱共享；2、针对共享频谱可用性的不确定性问题，提出等效信道容量新型评价指标和基于数据驱动的求解算法，用于指导在频谱可用性不确定情况下的有效频谱分配；3、考虑不同业务的多样化通信需求，针对频谱共享场景设计端到端的跨层资源规划方法，在频谱供应不确定情况下实现端到端的通信质量保证和高效率多维资源调度；4、提出基于深度强化学习的频谱感知和切换方法，实现高效、智能的动态频谱接入，提高对共享频谱的使用效率；5、面向6G空地一体化网络场景，将频谱共享技术与无人机网络深度融合，提出基于动态频谱接入的无人机中继流量卸载方案和边缘缓存方案，设计了基于深度强化学习的智能轨迹寻优方法，令无人机能够快速、自主地捕捉热点区域，并结合频谱环境自适应制定通信策略。前三项成果顺利完成本项目全部研究目标，后两项成果为本项目的后续拓展研究。综上，本项目超额完成目标，在高水平期刊和会议上共发表SCI、EI学术论文19篇（一作/通讯共12篇），其中IEEE Trans.等高影响力权威期刊论文11篇。申请国家发明专利4项，已授权2项。相关研究成果得到国内外众多学者的引用和认可，荣获辽宁省自然科学奖二等奖（排序第二）。