In a large road network with many signalized junction, signal timing optimization can be computationally expensive and its real-time implementationis questionable. To rectify this shortcoming, this paper is concerned with the concept of parallel computing as applied to both simulation model, Cell Transmission Model, and optimization model, Genetic Algorithm. Subnetwork division and parallel computing models are proposed and evaluated. Theachieved performance is reported with the basis on our severely congestednetwork, Sathorn Road, Thailand. Our study shows that this implementationis indeed computationally time-saving. Reduction of execution time depends on a number of subnetworks and simultaneous population computations.
Samenvatting