Comparative study of traffic assignment algorithms

An important problem in traffic planning occurs when trying to devise efficient methods to estimate Origin/Destination (OD) trip matrices from traffic counts in congested networks. The main objective of this research is to compare the performance of different traffic assignment algorithms in a model that estimates OD matrices. In particular, this study examines the relative efficiency of the traditional Frank-Wolfe algorithm and a new technique based on ‘Social Pressure’. These two algorithms are tested in a small network with two origins and two destinations. Results show that the Social Pressure based algorithm can perform better than the traditional approach leading to an improved efficiency of the matrix estimation model.