Mobile Subscriber Location Estimation based on Logistic Regression for Enhancement of Paging Procedure in 5G and beyond Networks

Mobility management in mobile networks is challenging, especially when the number of UEs is numerous and the moving patterns are diversified. It demands fresh approaches to minimize signaling overhead and maintaining service quality requirements. Mobile network operations generate vast data, and exploiting insights from big data with machine learning techniques is a strategy for operators to improve the efficiency of the operational system. This work introduces the logistic regression algorithm to predict the states of User Equipment (UE) based on its movement history and the paging policies adapt with the UEs states. The mechanism is proposed under the thorough investigation of data logs of mobile users operating data services of a mobile service provider in Vietnam. Evaluated with the data set of real mobile users, the results illustrate the improvement in the paging process, the participation of radio base stations in the paging procedure can be reduced by approximately 95%, while still maintaining a paging success ratio of over 95%, which is the required level of the mobile service provider.