A Network and Computing Resource Management Method based on Population Prediction in Mobile Networks

Existing traffic prediction methods are based on previously collected traffic patterns. The measured data is used to train and create a model, and the model predicts future traffic patterns. However, mobile traffic accounts for the majority of network congestion, and it is challenging to predict. Timely forecasting of exact mobile traffic is complicated due to complex spatio-temporal patterns of user demand. Even though we can obtain an accurate estimate of population in a target area, the number of active users cannot be accurately determined, and the traffic generated from a target area cannot be accurately estimated. In this study, we aimed at reducing the gap between population estimates, and the number of active mobile users in an area. Here, we have proposed a Bayesian model to represent the relationship between population and network and computing resources. The parameters that govern the Bayesian network are estimated from the collected data. We demonstrated that the proposed method reduced relative prediction error to 0.24, whereas the seasonal autoregressive integrated moving average (SARIMA) model had an error of 0.56.