AMC-SS Stochastic Modeling and Simulation for Traffic Flow

Proposal ID: 0606807

PI: Alexandros Sopasakis

Institution: University of Massachusetts Amherst

Title: AMC-SS Stochastic Modeling and Simulation for Traffic Flow

Abstract

Complex dynamical systems, such as traffic flow, are characterized by coexisting phases acting at different temporal scales. Further, random perturbations, although small, can have dramatic effects on the overall system behavior. While many physical phenomena result from the coherent but a priori unknown motion of a large number of variables most of the current models require that individual motion of the particles is resolved. Such approaches lead to difficulties in obtaining predictions that are dependable or timely. We propose a hierarchy of traffic flow models that can provide a better way of predicting traffic flow observables. Specifically we employ a novel, energy driven, stochastic noise model in order to describe the different complex phases of vehicular traffic flow. The model will be capable of reproducing the flow of traffic for the chaotic and difficult to resolve higher concentration regime.

Modeling the behavior of vehicular traffic flow is a challenging, current and open problem demanding an immediate solution. The proposed traffic flow model can be a breakthrough for several cross-disciplinary traffic related research areas by enabling much more accurate and potentially faster simulation of realistic traffic streams containing a large number of interacting vehicles. The coherent description of traffic can lead to better prediction strategies with direct and indirect benefits in our energy resources, environmental and safety issues.