Static Traffic Assignment on Ensembles of Synthetic Road Networks

We present a systematic approach for studying how performance of road networks is affected by changes in their geometry. We develop a new family of random planar graphs that models road networks and interpolates between a square grid and the β-skeleton of uniformly random points. The capacities of streets are set according to a rule that models a fixed provision of total resources. Ensembles of graphs are generated for different geometric parameter choices and the static traffic assignment problem is solved for a range of traffic demands. We find that variations in network efficiency, measured by the price of anarchy, are small both across demand values and geometric parameters. However, the best-performing networks are those which preserve some grid structure. We find that the price of anarchy does not correlate well with standard network statistics.