Predicting malfunction of mobile network base station using machine learning approach
Yin-Hsin Liu, Yao-Chung Tu, Chang-Yu Hsu, Hsin-Chieh Chao · 2019
In order to improve the communication quality of the mobile network, reduce the probability of malfunctions in the network components, reduce the time of service interruption and the labor cost of performing inspection and repair, the base station malfunction prediction mechanism is proposed, and it is expected that before the malfunction of the equipment occurs, it can be accurately predicted by the maintenance system, and then authorized dispatched units to conduct preventive malfunctions to examinations and repair. This article will describe an intelligent mobile network management system through the supervised learning of machine learning to achieve the purpose of intelligent maintenance. The system uses the alarm information of the base station received daily to combine with the malfunction dispatching information of the base station to complete the association between the time of occurrence of the malfunction and the alarm information of the K days before the occurrence of the malfunction. Through the Ensemble Learning to complete the establishment of the predictive model, and use the established model to predict the probability of a specific malfunction in the next M days. In the analysis of prediction results, precision and recall are used as indicators to measure the training model. It has been found through experiments that this model has a high probability of prediction and good performance. This prediction model has been applied to the base station malfunction prediction of the company's mobile network.