Accurate modeling of a large number of heterogeneous TCP flows is important for the understanding and control of Internet traffic. Difficulties in deriving such models arise due to the interaction between different protocol layers and a large state space size.

We introduce a stochastic model of a bottleneck ECN/RED gateway undera large number of competing heterogeneous TCP flows. Our main resultshows that as the number of flows becomes large, the queue dynamicsand the aggregate traffic are simplified and can be accurately describedby simple statistical recursions. These recursions can be evaluatedindependently of the number of flows, and hence the resulting trafficmodel is scalable. We also present a simple analysis on the bufferutilization and window size at the steady-state. Simulation resultssupporting the theoretical findings are also presented.