Modeling RTT Using Box-Jenkins Model

The Round Trip Time (RTT) is composition of propagation delay, transmission delay, queuing delay, router processing overhead and random delay due to medium access contention. The transmission delay, router processing overhead and propagation delay are deterministic component. The queuing delay and random delay are random noise component. The sudden and sharp increases of RTT caused by congestion, link failure and changes in routing configuration arising from heavy data traffic are random and difficult to model. Many models have been proposed in the literature to forecast the RTT. In this paper, we propose AutoRegressive „Integrated‟ Moving Average model (ARIMA(p,d,q) where p, q, d are the order of autoregressive process, moving average process and order of difference respectively. This model was popularized by Box-Jenkins for time series analysis. The ARIMA(p,d,q) model is a low pass filter which preserves the slowly varying trend component of a time series and removes the rapidly fluctuating or high frequency component. We found that ARIMA (2,1,0) is suitable for modeling the RTT. Most of the time, the one step forecast of RTT is unbiased with actual RTT.