Network Traffic Prediction Of Mobile Backhaul Capacity Using Time Series Forecasting

Abstract

Telecommunication tower company provides Mobile Backhaul service to provide end to end solution from base station to customer’s core network. This case study is conducted in one of the telecommunication tower company and mobile backhaul services provider that provides fiber optic connections as physical interfaces and ethernet transport equipment to serve the customer. Customer use leased line capacity mechanism to provide their requirement on mobile backhaul connectivity. The bandwidth capacity may encounter an increase in daily or monthly usage, which requires the customer to upgrade their maximum capacity. As a service provider, PT Tower Bersama wish to predict the customer bandwidth utilization to discern when the customer needs to upgrade their mobile backhaul leased line capacity. The network traffic is modeled as a time series data. Fractionally Auto Regressive Integrated Moving Average (FARIMA) model and Artificial Neural Network (ANN) model are used to forecast the future network traffic. In terms of error FARIMA (4,0.2,1) model shows the least error with RMSE, MAE and MAPE are 11.762, 9.329 and 11.950 respectively. However, ANN-MLP model prediction result shows more similar pattern with the existing traffic with a slight difference in error with FARIMA model. The prediction model then applied to the interactive dashboard to determine client’s upgrade based on the forecasted traffic data.