Scaling Up Transit Priority Modelling Using High-Throughput Computing

Mesbah, M., Sarvi, M., Tan, J. and Karimirad, F.

The optimization of Road Space Allocation (RSA) from a network perspective is computationally challenging. An analogue to the Network Design Problem (NDP), RSA can be classified NP-hard. In large-scale networks when the number of alternatives increases exponentially, there is a need for an efficient method to reduce the number of alternatives while keeping computer execution time of the analysis at practical levels. A heuristic based on genetic algorithms (GAs) is proposed to efficiently select Transit Priority Alternatives (TPAs). The proposed framework allows for a TPA to be analysed by a commercial package that is a significant provision for large-scale networks in practice. We explore alterative parallel processing techniques to reduce execution time: multithreading and High-Throughput Computing (HTC). Speedup and efficiency are compared with that of traditional sequential GA, and we discuss both advantages and limitations. We find that multithreading is better when using the same number of processors, but HTC provides expandability.

Cite as: Mesbah, M., Sarvi, M., Tan, J. and Karimirad, F. (2012). Scaling Up Transit Priority Modelling Using High-Throughput Computing. In Proc. Australasian Symposium on Parallel and Distributed Computing (AusPDC 2012) Melbourne, Australia. CRPIT, 127. Chen, J. and Ranjan, R. Eds., ACS. 53-62

(from crpit.com) (local if available)