面向铁路应急通信的认知自组网分簇协作频谱感知

With the rapid development of railway, the higher requirement for the capability of emergency communication is put forward. However, the less service types, slowly available time, complicated network deployment and limited spectrum resource in current railway emergency communication network have caused low efficiency. Therefore, it necessitate a new railway emergency communication paradigm, referred to as clustering cognitive ad hoc network, in order to achieve high emergency efficiency. An effective clustering index set will be built up using Petri Net theory, and a reasonable weight allocation for the indexes that can adaptively vary with the railway emergency communication scenarios is explored. Besides, the maintenance of clusters with low overhead and high stability is designed by Monte Carlo optimization. Moreover, a cooperative spectrum sensing algorithm incorporated with clustering is investigated, in which the cooperative condition, the cooperative mode, and the influencing factors on the cooperation are obtained to improve the efficiency of the railway emergency communication. Further, a hierarchical strategy for the common control channel is presented in the case of different cooperative modes for intra and inter clusters communication in the spectrum sensing. The goal of the project is to achieve a promising cluster-based cooperative sensing algorithm for railway emergency application, with the performances of high stability, strong time-effective and short sensing time (in the order of micro-second). The achievements of the project will provide theoretical foundations and technical support for a novel and efficient network deployment in railway emergency communication scenario.

铁路建设的飞速发展对其应急通信保障能力提出了更高要求。针对目前铁路应急通信网络因业务功能少、开通时限长、网络部署复杂和频谱资源紧张而导致的通信效率低下问题，本项目研究用认知自组织（Ad hoc）分簇组网方法构建铁路应急通信系统。拟借助Petri网理论建立有效的分簇算法指标集；研究能提高权重分配合理性、自适应应急通信场景变化的指标权重分配算法；用蒙特卡洛优化理论设计具有低维护开销和高稳定性的簇维护机制；研究能进一步提升铁路应急通信效率的分簇协作频谱感知算法，掌握分簇与频谱感知相互协作的条件、方式以及影响协作程度的因素；研究分簇协作式频谱感知算法中，簇内、簇间等采用不同协作方式时的分级公共控制信道设计机制。旨在提出高稳定性、感知时长为微秒（us）级、时效性强的铁路应急认知Ad hoc网络的分簇协作频谱感知新算法，为高效的铁路应急通信组网实现提供理论基础和技术支撑。

铁路建设的飞速发展对其应急通信保障能力提出了更高要求。针对目前铁路应急通信网络因业务功能少、开通时限长、网络部署复杂和频谱资源紧张而导致的通信效率低下问题，本项目对认知的铁路应急分簇组网模型及协作频谱感知算法展开了研究。主要研究内容为：1）铁路应急认知Ad hoc网络加权分簇算法的考量指标。仿真平台以青藏线格尔木站-南山口站为模型，建立了以GSM-R终端节点度、节点移动性、节点能量以及终端功能等级分簇算法考量指标集。2）基于加权分簇的自适应指标权重分配算法。利用NS2仿真工具对分簇的铁路应急单呼及组呼业务的仿真结论为：当应急节点具有较低移动性时，平均端到端时延维持在0.5s以下，语音和数据业务指标要求均能满足GSM-R指标要求；当节点移动速度大于50m/s时，时延值仅能满足数据业务指标要求。3）簇维护机制的研究。利用蒙特卡洛优化理论解决了节点权值频繁抖动所致的簇头频繁替换问题。仿真结论为：平均簇头数目N随着节点传输距离的增加而减少；N随节点速度的改变发生频繁的波动且波动幅度较小；提出的簇维护算法具有较少的簇头更新次数。4）基于分簇的协作频谱感知方案及算法。仿真结论为：在虚警概率值一定时，平均感知比特开销K有了较大程度的降低；允许一定的虚警概率时，漏检概率随K的变化分为两种情况：①K<0.5时，漏检概率随K的增加有非常显著的下降；② 0.5≤K≤1时，漏检概率随K的增加变化较为缓慢。这说明，漏检概率性能的略微牺牲便可换取近50%平均感知比特开销的降低。5）分簇协作频谱感知算法的公共控制信道设计机制。在研究内容4中采用故障小区的1对载频用作协作感知的公共控制信道，有效降低了CR节点的信道接入时延和感知频谱的频繁切换开销。本项目提出了一种高稳定性、具有较低感知开销（小于1bit）、较少感知时长（us级）、时效性强的铁路应急认知Ad hoc网络的分簇协作频谱感知新算法，为高效的铁路应急通信组网实现提供了理论基础和技术支撑。