Distributed Computing Approaches for Large-Scale Probit-Based Stochastic User Equilibrium Problems

Traffic assignment on transportation networks is often accomplished using determinant user equilibrium (DUE) or stochastic user equilibrium (SUE). However, these applications can be limited because of the heavy requirements for computing power required. This article presents a distributed computing approach that uses a two-stage Monte Carlo simulation method to solve the stochastic network loading problem. The authors’ work accelerates the Monte Carlo simulation method by using a distributed computing system. They adopt three distributed computing approaches for the workload partition of the Monte Carlo simulation method. The first approach allocates each processor in the distributed computing system to solve each trial of the simulation in parallel and in turns. The second approach assigns all the processors to solve the shortest-path problems in one trial of the Monte Carlo simulation concurrently. The third approach combines the first two, so different trials of the Monte Carlo simulation as well as the shortest path problems in one trial are solved simultaneously. The authors report on their comprehensive testing of these three approaches, first on the Sioux-Falls network and then on a randomly generated network example. Their results demonstrate that computational time for the probit-based SUE problem can be largely reduced by any of these three approaches. However, they stress that the first approach is the best one, which they demonstrate by using it to calculate the probit-based SUE problem on a large-scale network example. These techniques may prove useful in areas such as road pricing toll design and transportation network design problems.


  • English

Media Info

Subject/Index Terms

Filing Info

  • Accession Number: 01504633
  • Record Type: Publication
  • Files: TRIS
  • Created Date: Jan 24 2014 2:30PM