Prediction of Network Throughput using ARIMA

In this paper, we apply an ARIMA (Autoregressive Integrated Moving Average) model to predict future network throughput, which is important in improving the network protocols in terms of latency, energy, etc. The Autoregressive Integrated Moving Average (ARIMA) model is a popular and a successful method to predict time-series data. It has wide applications in time-series analysis, including statistics and economics. We first train the model with the network throughput data using history. Then we make an estimation for the throughput in the future. We use the Mean Squared Error (MSE) as a means of error estimation and tune the parameters $p, d, q, m$ of the ARIMA model. As a result, we obtain a forecast waveform with an average error rate of 2.84%.