Parallel implementations of dynamic traffic assignment models and algorithms for dynamic shortest path problems
Author(s)Jiang, Hai, 1979-
Massachusetts Institute of Technology. Dept. of Civil and Environmental Engineering.
MetadataShow full item record
This thesis aims at the development of faster Dynamic Traffic Assignment (DTA) models to meet the computational efficiency required by real world applications. A DTA model can be decomposed into several sub-models, of which the most time consuming ones are the dynamic network loading model and the user's route choice model. We apply parallel computing technology to the dynamic network loading model to achieve faster implementations. To the best of our knowledge, this concerns the first parallel implementations of macroscopic DTA models. Two loading algorithms are studied: the iterative loading algorithm and the chronological loading algorithm. For the iterative loading algorithm, two parallelization strategies are implemented: decomposition by network topology and by time. For the chronological loading algorithm, the network topology decomposition strategy is implemented. Computational tests are carried out in a distributed-memory environment. Satisfactory speedups are achieved. We design efficient shortest path algorithms to speedup the user's route choice model. We first present a framework for static shortest path algorithms, which prioritize nodes with optimal distance labels in the scan eligible list. Then we apply the framework in dynamic FIFO, strict FIFO, and static networks. Computational tests show significant speedups. We proceed to present two other shortest path algorithms: Algorithm Delta and Algorithm Hierarchy. We also provide the evaluations of the algorithms.
Thesis (S.M.)--Massachusetts Institute of Technology, Dept. of Civil and Environmental Engineering, 2004.Includes bibliographical references (p. 139-144).
DepartmentMassachusetts Institute of Technology. Dept. of Civil and Environmental Engineering.
Massachusetts Institute of Technology
Civil and Environmental Engineering.