Terminal positioning method and system based on machine learning

Abstract

The invention discloses a terminal positioning method and system based on machine learning. The terminal positioning method comprises the step of: acquiring position measurement information which corresponds to a terminal and comprises MDT data, and carrying out the feature extraction of the position measurement information to obtain position feature information; based on a preset algorithm, training to obtain a position prediction model according to the position feature information; and acquiring position measurement information of the to-be-positioned terminal, and inputting the position measurement information of the to-be-positioned terminal into the position prediction model to obtain a longitude and latitude positioning result of the to-be-positioned terminal. According to the terminal positioning method, massive user data in the current network is acquired, a position prediction model is obtained by training in a big data mode, the terminal is positioned by using the position prediction model, the positioning method is directly learned from the massive data of the current network and is used for positioning according to the data of the current network, that is to say, the model itself fully considers the wireless propagation environment and link reflection characteristics of each positioning point, and accurate positioning can be realized.

Description

technical field [0001] The present invention relates to the field of wireless technology, in particular to a machine learning-based terminal positioning method and system. Background technique [0002] With the rapid development of wireless communication technology and the popularity of smart terminals, services based on the location of end users have shown explosive growth. Accurately locating the location of the terminal is of great significance for providing users with better basic services and adding richer value-added services. However, the wireless propagation environment is becoming more and more complex. The refraction, reflection and fading environment in the wireless signal propagation are randomly distributed, which poses a severe challenge to the accuracy of positioning. More advanced and accurate positioning methods need to be found. [0003] Among the existing terminal positioning methods, the most common terminal position positioning methods include a three-p...

AI Technical Summary

Problems solved by technology

These two methods have the following disadvantages: 1) It is necessary to calibrate the radio wave propagation model for each site, which is heavy workload and huge cost; 2) The calibration of the propagation model is only a statistical model, which is applied to each actual end user There is a large error in positioning; 3) For the delay of radio waves, affected by the relay components in the transmission path and the propagation path of radio waves in the air, the delay cannot truly reflect the straight-line distance from the user to the base station antenna, so it is often used to calculate the distance using the delay Inaccurate; 4) 3G and 4G technical characteristics, the system only measures the delay from the terminal to a main service cell, that is, only one point of distance can be obtained, and it is impossible to form a three-point positioning based on delay, and the distance to the other two points remains the same. It needs to be calculated from the aforementioned radio wave loss method

[0005] In the fingerprint identification method, the fingerprint library data mainly comes from road test data, and the acquisition of data is time-consuming and laborious, and the data volume is difficult to reach the level of massive data; at the same time, its data source is limited by the route and range of the measurement travel, and the data source is a line segment The data on the network plane has relatively large positioning limitations

[0006] It can be seen that the traditional positioning method based on the terminal measurement report requires propagation model calibration, which is heavy in workload and costly, and the positioning accuracy is not high due to the loss of radio wave delay.

However, the fingerprint identification method has a single data source and is affected by the route and range of travel, so the positioning is limited.

Images