As the most basic resource in mobile communication technology,the wireless spectrum resource plays an important role in mobile communication.With the coming of the Fifth Generation(5G)era of mobile communication technology,the wireless spectrum resources are becoming more and more complex and difficult to manage,making the utilization rate of wireless spectrum resources always at a low level.Therefore,more efficient timefrequency resource allocation algorithms and technologies need to be applied urgently.In the current field of Railway wireless communication,the GSM for Railway(GSM-R)communication system based on the Global System for Mobile Communication(GSM)platform is evolving to a new generation of High-Speed Railway mobile communication system(LTE-R).Under the premise of ensuring High-Speed Railway’s(HSR)operation and safe communication,it can provide users with higher bandwidth and lower delay services.How to make full use of the limited wireless spectrum resources in railway communication system and improve the utilization of wireless spectrum under the premise of reliable communication for HSR is the main research direction of railway communication system at present.In addition,with the coming of the concept "spectrum situation",it provides a new idea for the research on the fully use of railway communication wireless spectrum resources.This thesis based on high speed railway intelligent spectrum as the main research content,through the analysis of HSR communication scenarios’ users service network selection,cell handover and channel selection these three problems,respectively studied the opportunistic access situation of the spatial dimension,neighborhood handover situation and time dimension’s channel occupancy situation.By combining the Graph Convolutional Network(GCN),Reinforcement Learning(RL)theory and temporal depth neural network,respectively studied and modeled spectrum situation of different dimensions,proposed the·new wireless spectrum resource optimization algorithms,and with the communication simulation,demonstrated the studies has the trend of validation applications,and improved the utilization of the wireless spectrum under the HSR communication environment.The main innovations of this thesis include the following three points:Firstly,aiming at the problem of cell selection and cell access of HSR communication in heterogeneous Network,the spatial characteristics of intelligent spectrum situation are studied,and the opportunistic access situation of railway communication system based on the GCN is established.By using GCN network to mine the spatial correlation between different cells,combining with the traditional heterogeneous network multi-attribute decision algorithm,a railway communication community selection algorithm based on GCN network was proposed to further improve the rationality of cell selection.Secondly,aiming at the parameter selection problem of cell handover in HSR communication system,the cell handover situation in smart spectrum situation was studied,and a parameter selection algorithm based on reinforcement learning and parameter estimation is established.The algorithm adopted the interaction between agent and environment,the handover performance and network performance under different handover parameters were obtained under dynamic railway environment conditions,and selected the handover parameters with the best communication performance to improve the handover success rate and service quality of train users.Finally,in view of the HSR users in the channel selection problem,an intelligent spectrum situation was studied based on the time features of channel occupancy situation,using sequential depth neural network for authorized channel utilization situation awareness,according to the authorized channel of time-series data to predict the future’s changes,enhance the accuracy of channel occupancy situational awareness.It provided a reference for users to access the authorized spectrum dynamically and improved the utilization rate of the wireless channel spectrum resources. |