Localization of a high-speed train using a speed model based on the gradient descent algorithm

Owing to the problem of weak coverage of private network in the high-speed rail (HSR) environment, the users' internet perception will be seriously affected by abnormal connection to the ordinary public network. Radio remote unit(RRU) that works abnormally is an important reason for the weak coverage of HSR private network. To address this problem, first, in contrast with the traditional positioning system, such as the global system for mobile communications-railway(GSM-R), we proposed a method that just use HSR user's information collected by base stations and base station information provided by the telecom operators of China to locate a high-speed train(HST) off-line. Second, an abnormal RRU was detected by using the established speed models based on the gradient descent algorithm. The experimental results showed that the probability of accurately locating the HST was up to 87.17%.