Reducing queue-fill variability for emergency traffic in a differentiated services network

Supporting emergency services applications using the Internet has gained considerable interest in the recent past. A simple multiple average multiple threshold (MAMT) random early detection (RED) scheme is used to provide the necessary quality of service for emergency traffic. We have studied the RED dropping function as one of the contributing factors towards the variation observed in the queue fill. This may cause interactive emergency applications to experience large delay and delay variation thus deteriorating the service provided to them. Hence, we compare the performance of the linear dropping function in terms of average delay and variation in queue fill with that of the three other dropping functions namely: concave, convex and step. With the help of an exhaustive simulation study conducted using the network simulator-2, we conclude that the concave function provides the best performance