Global and Local Information in Traffic Congestion

TL;DR

This paper compares global versus local information control protocols in a network flow model, revealing trade-offs between flow capacity and fluctuations, and explaining the emergence of pulsating jams.

Contribution

It introduces a generic network flow model analyzing the impact of global and local information protocols on congestion dynamics.

Findings

Global information protocol allows higher flow before jamming.

Local protocol results in smaller flow fluctuations.

Jamming propagates as pulsating cores in the network.

Abstract

A generic network flow model of transport (of relevance to information transport as well as physical transport) is studied under two different control protocols. The first involves information concerning the global state of the network, the second only information about nodes' nearest neighbors. The global protocol allows for a larger external drive before jamming sets in, at the price of significant larger flow fluctuations. By triggering jams in neighboring nodes, the jamming perturbation grows as a pulsating core. This feature explains the different results for the two information protocols.