Time-dependent collective behavior in a computer network model

The collective dynamics of large-scale computer networks remains elusive due to not only the internal adaptive behaviors of network-wide flows, but also the spatial–temporal changes in the external environment. In this paper, we investigate the time-dependent collective behavior by using a computer network model, recently developed to study space–time characteristics of congestion in large networks. We use the evolving correlation pattern, the largest eigenvalue, and the information entropy to analyze the macroscopic pattern of changing network congestion. We find the collective behavior becomes more pronounced during transient periods of pattern shifting, and the macroscopic pattern becomes gradually indistinct as the observed timescale increases to some extent. We also find that the evolving pattern of spatial–temporal correlation is more useful to reveal the time-dependent collective behavior of our model at different forcing levels.