Deep Learning based Mobile Traffic Flow Prediction Model in 5G Cellular Networks

Abstract

Due to drastic development of smart phone devices, there is a considerable increase in mobile traffic flow that is provided difficulty on the fifth generation (5G) cellular networks. The growing quantity of communication data along with the need for maximum peak data rate, enhanced reliability and minimum delay in 5G network poses several challenging issues. At the same time mobile traffic flow prediction in 5G network is a challenging topic. In this aspect, this paper proposed a novel DL based mobile traffic flow prediction (DLMTFP) model in 5G networks. The goal of the DLMTFP technique is to differentiate the user's mobile traffic, application usage and traffic patterns. In addition, the DLMTFP technique involves the modelling of bidirectional long short term memory (BiLS TM) model for the estimation of the mobile traffic flow. Moreover is employed for the hyper parameter optimization of the BiLSTM model. The exploitation of the hyper parameter optimization process helps to significantly boost the predictive outcome of the BiLSTM model. The DLMTFP technique is able to predict the traffic flow in peak range. For validating the improved predictive performance of the proposed DLMTFP technique a comprehensive simulation analysis takes place over the other state of the techniques.