Performance Evaluation Of High-Speed Railway Wireless Network Using Stochastic Network Calculus

In recent years, to satisfy the large amounts of data requirements and strengthen railway operation safety guarantees, Long Term Evolution for Railway (LTE-R) is put forward to be the next generation wireless communication for High-speed Railway (HSR). In order to better guarantee safety and reliability of railway, the precise analysis of system performance should be carried out before the operation of HSR, accordingly, meeting a certain performance target after optimization. LTE-R, the application of LTE (the expectant4th generation technology) into HSR scenario, succeeding the key technologies of LTE, such as employing all-IP packet switching, which means all the original and novel traffic will be transmitted through packet switching pattern.In the past, the previous research of performance analysis of HSR mainly concentrated on programming simulation method. There is no doubt that these methods can analyze performance relatively precisely, however, there also exits limits to their application, on account of longer time consumption and higher cost. However, stochastic network calculus (SNC), a new theory tool of performance analysis in networks, is advancing rapidly, solving the performance guarantees in modern packet-switched network. With less time consumed and lower cost, it is easier to implement and becomes a new way to analyze the network performance of LTE-R system. This paper focuses on HSR network performance analysis utilizing stochastic network calculus theoretical method, involving modeling specific traffic model in HSR based on practical measurement data and modeling the channel taking large-scale fading and small-scale fading into account. Besides, this paper makes verification about the accuracy of SNC and analyzes network performance such as delay in HSR scenario according to the results achieved. The innovations of this paper are listed as follows:First, the precise packet-level Markov model which takes into account the channel variations due to both large-scale and small-scale fading effects has been constructed. The mobility model of high-speed train is formulated as a semi-Markov process, so that the instantaneous data rate of wireless channel becomes a semi-Markov modulated process. Moreover, the performance loss due to AMC selection with imperfect CSI is also considered. Then, the stochastic service curve of HSR communications system is derived based on the semi-Markov modulated process.Second, the precise and appropriate traffic model for specific traffic in high-speed scenario has been constructed. The stochastic arrival curve of train control services are proposed based on practical measurement data.Third, both CCDF SNC and MGF SNC approach are used to derive the delay and backlog bounds of train control services and the results are compared.Last, the analytical delay bound is validated by simulation and can be used in the design and dimensioning of LTE-R system. Besides, the next generation high-speed railway communication system LTE-R has been evaluated based on the derived delay bound.