Publication Details
Cellular automata based traffic simulation accelerated on GPU
Sekanina Lukáš, prof. Ing., Ph.D. (DCSY FIT BUT)
Fučík Otto, doc. Dr. Ing. (DCSY FIT BUT)
Traffic, Microsimulation, Cellular automata, Acceleration, GPU.
This paper presents acceleration of traffic microsimulation model based on cellular automata. Program is implemented using NVidia CUDA SDK. Various implementations were tested.
Intelligent transportation systems become more and more important with the increasing traffic densities and safety requirements. A reasonably good traffic prediction can be obtained using microscopic traffic simulation models witch distinguish and trace every traffic entity. However, microscopic simulation requires considerable computing resources. In this paper, we propose to accelerate a cellular automata based microscopic traffic simulator using graphic processing units (GPU). The proposed accelerator provides speed-up of 204.65 with respect to a single core solution for problems instances containing 170 mil. cells, equivalent to 935 000 km of traffic network. This solution is sufficient to predict traffic simulations multiple in real-time.
@INPROCEEDINGS{FITPUB9555,
author = "Pavol Kor\v{c}ek and Luk\'{a}\v{s} Sekanina and Otto Fu\v{c}\'{i}k",
title = "Cellular automata based traffic simulation accelerated on GPU",
pages = "395--402",
booktitle = "Proceedings of the 17th International Conference on Soft Computing (MENDEL2011)",
year = 2011,
location = "Brno, CZ",
publisher = "Institute of Automation and Computer Science FME BUT",
ISBN = "978-80-214-4302-0",
language = "english",
url = "https://www.fit.vut.cz/research/publication/9555"
}