Machine Learning Based Mobile Data Traffic Prediction in 5G Cellular Networks

Abstract

The rapid evolution of cellular technologies has resulted in a drastic increase in mobile data traffic. Particularly, in 5G cellular networks, the design of accurate time-series models become essential to predict and improve the mobile data traffic and quality of services (QoS). The mobile data traffic prediction models allow the operators to adapt to the traffic demands of the network with improved resource usage and user experience. In addition, the prediction of mobile data traffic is a tedious process due to the nature of high heterogeneity amongst distinct base stations with varying traffic loads. Therefore, several artificial intelligences (AI) based machine learning (ML) and deep learning (DL) models have been developed for mobile data traffic prediction. This paper provides a comprehensive review of existing ML models to predict mobile data traffic in 5G networks. Moreover, the existing techniques are reviewed based on different aspects such as major objectives, underlying methodology, advantages, inferences, and performance measures. An extensive comparati ve study of the surveyed approaches also takes place to identify the unique characteristics of every technique. Finally, a summary of challenging issues and future directions are discussed in detail.