Abstract
Understanding and evaluating TCP behaviour remain a challenging problem, because of the complexity of the protocol itself and the inherent complexity of the interactions between the protocol and the network. Several analytical models of the TCP behaviour have been proposed recently. Most of these models assume a fixed number of persistent TCP connections. A few of the recent studies on TCP look at non-persistent TCP connections but with many limiting assumptions. Moreover, nearly all non-persistent studies are macroscopic and not scalable. This paper introduces a scalable stochastic reward Petri nets (SRNs) model for non-persistent TCP sessions that share two bottleneck links in a wide area network (WAN). The suggested model is constructed in a microscopic approach which captures the essential protocol features of the congestion control mechanism used by the TCP Reno protocol. Several limiting assumptions have been relaxed by this model. The interference from other non-persistent sources is introduced to the model in an efficient way. Several performance metrics have been captured and compared to simulation results.