Offline estimation of origin-destination matrices using simultaneous perturbation stochastic approximation algorithm

Abstract

The origin-destination (OD) matrix is the important input in many trans- portation planning problems. Recently various tra±c simulation softwares has become available, which made it possible to perform the assignment of OD ma- trix to transportation network in more e±cient way. Use of simulation brought up a need for e±cient simulation optimization techniques, that can be applied to the OD matrix estimation problem. Recent research showed some inspiring results on the performance of Simultaneous Perturbation Stochastic Approxima- tion (SPSA) alorithm. The aim of this research is to justify the use of SPSA for OD estimation, test the performance of the algorithm and develop the modiЇca- tion that can improve it.