Researches On Time Distribution Model And Handover Decision Algorithms In GSM-R System

Under rapid development in Chinese railway, all existing wireless systems for railway are generally not suitable for requirements in High Speed Railway, Passenger Dedicated Line, etc. GSM-R system is the best choice for requirements of railway applications, which is based on GSM that is the one of maturative techniques in public wireless systems. Compared with GSM system, it is added to GSM-R system that voice group call service, voice broadcast service, function address, position address, etc.In public wireless systems, the cell residence time, call holding time, channel occupancy time of the user are hot research points which are essential for system designs and system performance evaluations. So, the cell residence time, call holding time, channel occupancy time of the train are also important in GSM-R system, although GSM-R system is dedicated for railway. Moreover, for affording continuous communication when the train is running highly, the handover decision algorithm, which must make the average number of handover minimal, is the one of most essential techniques in GSM-R. Therefore, both time distribution model and handover decision algorithm in GSM-R system are main tasks in this paper.For time distribution model, a novel random number generation method for residual life time is firstly proposed which is based on the acceptance-rejection method. And, it is suitable for the case where the average value of cell residence time is converged to the finite value. Secondly, an analytical call holding time model is proposed in which both cell residence time and the average number of cells, which the user undergoes during his call that is unencumbered, are only concerned on. Thirdly, depending on Provisional Design Specifications for Passenger Dedicated Line, a method for generating the length of the line in Section of Passenger Dedicated Line is proposed. Then, considering two different cell coverage methods, the length distribution within GSM-R cell in Section of Passenger Dedicated Line is proposed. Furthermore, train cell residence time, train call holding time, train channel occupancy time in Section of Passenger Dedicated Line are modeled one by one. Finally, according to both Provisional Design Specifications for Passenger Dedicated Line and train technical operations in Station of Passenger Dedicated Line, the train cell residence time model for No-Stop-Train and Stop-Train is proposed within GSM-R cell covering Station of Passenger Dedicated Line.For handover decision algorithm, considering the train having the certain running direction when it is running in Section of Passenger Dedicated Line, the novel handover decision algorithm is firstly proposed which has a random-direction lock and encourages the handover towards the running direction of the train. And, the proposed algorithm could reduce the average number of handovers efficiently, up to 25%. Secondly, an adaptive handover decision algorithm is introduced into GSM-R system, which performs well in public wireless systems. Then, using Genetic Algorithm, two key parameters of the adaptive handover decision algorithm are optimized for fitting the scenario in GSM-R system. From results obtained by a lot of optimizations, two key parameters of the adaptive handover decision algorithm are sensitive to the velocity of the train.