Using AR(I)MA-GARCH models for improving the IP routing table update

When an IP router updates its routing table, e.g., upon failure detection, network traffic is lost as long as the routing entries affected by the failure are not updated. In this paper, we model and predict the network traffic passing through an IP backbone router and define a dynamic heuristic in order to reduce the packet loss resulting from such routing tables entries update events. We use the state-of-the-art AutoRegressive Integrated Moving Average (ARIMA) - Generalized AutoRegressive Conditional Heteroskedasticity (GARCH) model to characterize and predict the network traffic. We benchmarked the resulting models with a heuristic in a simulation environment and show that a significant decrease of packet loss can be obtained for a given computational cost.