Research On Importance Sampling Technology Of Communication System In High-Speed Railway

As the high-speed railway's information bearing platform, the railway digital mobile communication system GSM-R has been developed rapidly, and is becoming mature. To ensure traffic safety, and further to meet the passenger's demand for communications services, high-speed railway has a higher demand on the GSM-R system's reliability, availability, maintainability and safety. In engineering practice, all the indicators of network quality would be tested to judge their compliance with EIRENE's functions and system requirements specification. The procedure will be both in early stage of the construction of GSM-R networks and in network optimization stage after the construction. Modern research on GSM-R theory more often studies the radio wave propagation under high-speed railway scenes, interference analysis and wireless channel modeling through the laboratory's simulation platform, and has less attention to the bit error rate (BER), which most directly reflects the reliability of communication system. This is because the BER of GSM-R system is required to be below10-6. For low BER estimating, traditional BER estimate methods, such as the Monte Carlo as the representative method, would need large number of samples, more time consuming. And these methods can not achieve accurate estimate on the BER of GSM-R system. In response to this situation, importance sampling algorithm is introduced to the research on system's BER, so as to reduce the simulation time with its advantages, and to estimate the low BER of GSM-R system.The background and significance of the research is proposed first, also a review for the domestic and international research on importance sampling in all kinds of fields are progressed. Then the wave propagation properties and wireless channel characteristics in the high-speed rail system environment is introduced. Then the factors, that impact the communication quality of GSM-R system and directly affect the BER, are studied. In order to estimate the transmit power with the research results of the BER, two commonly used wave propagation path loss calculation model are compared for the proper high speed railway scene. And the Hata model, which more suitable for the scene, is used to estimate the path loss for different scene.Then the importance sampling method is analysised and expounded both from the mathematical and communication aspects. Based on the analysis and conclude of the design methods for importance density function, the shortcomings of the existing design methods are pointed out. And then a new importance density function is proposed.An importance sampling algorithm for the BER estimation of GSM-R system is presented, and then the design of the importance density function is completed through several experiments. The experimental results show that the traditional Monte Carlo method would be more time-consuming and need large number of samples for low BER, or even the complexity is beyond the computer performance. While with the proposed importance sampling method in this paper, the simulation time required for estimate the same low orders of magnitude's BER can be greatly reduced. The timeliness of the simulation has been improved. Finally, the experimental data is applied with GSM-R system margin design to estimate the transmitting power for GSM-R system in different circumstances. Verified the practicability of estimate the transmit power from the BER perspective.