Particle-based and stochastic simulation-based traffic assignment

Abstract

The particle-based and stochastic traffic assignment problem is considered. The increasing deployment of “agent-based” traffic assignment models in transport planning renders this problem relevant. Lacking a comprehensive mathematical characterization, the computation of an assignment within this model class is bound to heuristics. One of these approaches is the “sorting technique” of Sbayti et al. (2007). The present article offe: derivation of this method from first principles, works out the effect of model stochasticity, and uses these results to propose a suitable simulation noise filtering rule. Experimental results covering a wide range of variations of the method under different congestion levels illustrate its advantageous performance.