Predicting the Load Capacity of 4G Cellular Networks With Deep Learning

Abstract

Predicting the capacity of cellular communication networks is an important factor to provide better services to subscribers. As the number of mobile subscribers increases, the network load and user experience increase. By predicting the channel quality indicator (CQI) as a main factor in the network performance and spectral efficiency, it is possible to check the experimental quality in terms of appropriateness for the desired environment. In this article, the authors aimed to investigate the performance of the mobile phone network capacity of Mobile Telecommunication Company of Iran (MCI) using CQI prediction employing deep learning methods. To increase the accuracy of the proposed deep network model, hand designed features such as frequency band, physical resource block (PRB), the number of surrounding cells within a radius of 2.5 km, download/upload payload, and modulation are extracted and fed as the model input. The proposed model can predict the CQI with 96% mean absolute error rate on the real dataset of cell stations.