SCSC2003 Abstract S566
Investigating Compressibility of Large-Scale Microscopic Transportation Systems Simulation
Investigating Compressibility of Large-Scale Microscopic Transportation Systems Simulation
Submitting Author: Dr. Sherif Ishak
Abstract:
Discrete system simulation continues to offer transportation modelers an effective and robust tool to model the stochastic operation of traffic flows. The significant advancements in computing powers over the last decade have motivated the development of more comprehensive and sophisticated microscopic simulation systems. Current simulation models are capable of simulating composite transportation networks that are almost impossible to model analytically. However, the computational requirements for simulating such systems can be extremely intensive for both off-line and on-line applications. The required computing resources for microscopic system simulation are known to increase exponentially with the size of the network and the number of entities (vehicles) to be simulated. In order to reduce the extensive computing requirements of large scale transportation systems, an approach is proposed and investigated in this paper to reduce the inherent redundancies in the simulated
system and thereby reduce the overall complexity. The approach attempts to scale down or compress the original network into a smaller network size, while maintaining the geometric and kinematic similitude. The compressed network can then be simulated much faster to gather the relevant performance measures, which can be converted back to the original model. By doing that most of the inherent redundancies in the original network can be reduced and the simulation performance can be dramatically improved. This paper presents an exploratory analysis to achieve this objective. The results are very encouraging at the preliminary stage and further research is underway.
Back to SCSC2003 Abstracts