Analysis And Application Of GSM-R Transmission Interference Based On Data Mining

With the developing of high speed railway like Beijing-Shanghai, Beijing-Shenzhen, Shanghai-Wuhan, and opening to traffic successively, the "four vertical and four cross" high speed rail network of China is gradually formed. To ensure the security and reliability of rail transportation system, CTCS-3system should provide continuous and seamless network services for board ground communications in all kinds of topography and areas along the railway line, and complete the transmission of critical information like the train position tracking, movement authority, emergency parking, temporary speed regulating and so on. However, many factors such as Doppler frequency shift, handoff and multipath effect will affect the transmission of train control information. Therefore, studying the transmission interference of the GSM-R network is of great significance.There are many factors affecting the GSM-R network transmission interference. In consideration of the GSM-R network itself, the main factors are the transmission interruption caused by handoff, the path loss of the field coverage and the signal interference from base stations with co-channel and adjacent-channel frequency. Besides, the Doppler frequency shift due to the high speed of the train is another factor. Moreover, the environment is also an important factor, while the complex terrain and bad weather can cause deterioration of the network. Through GSM-R network test, mass data of network will be produced. Based on the foundation of GSM-R data acquisition, network test, and distributed storage of data, this paper provides an improved SPRINT parallel decision tree data mining algorithm to do data analysis. And with the advantages of distributed parallel computing, we obtained a transmission interference prediction model effectively. At the same time, we use BP neural network algorithm and history field strength coverage data to build a GSM-R field strength prediction model. Through combining with the field strength prediction and the transmission interference prediction, we propose a method for the transmission of train control information, to predict the probability of transmission interference and control the send probability of data. Simulation results show that the method can significantly reduce the probability of data interference, improve the reliability of the train control information transmission, and be used in practical application.