Phase transition in a computer network traffic model

Abstract

We propose and study here a simple model of computer network traffic that can exhibit a phase transition from a low to high congestion state measured in terms of average travel time of packets as a function of the packet creation rate in the network. In the model, packets are generated with destination addresses, and are transferred from one router to another toward their destinations. The routers are capable of queuing packets and autonomously selecting a path to the next router for a packet. Through simulations on a two-dimensional lattice model network, we found that the phase transition point into the congestion phase depends on how each router chooses a path for the packets in its queue. In particular, an appropriate randomness in path selection can shift the onset of traffic congestion to accommodate more packets in the model network.