A distributed implementation of genetic algorithms for dynamic traffic routing

The development of dynamic traffic routing strategies that aim at improving the utilization of a transportation network capacity is a challenging task that involves solving a complex optimization problem. The current paper explores the applicability of using a distributed implementation of Genetic Algorithms (GA's) for solving the problem, which allows optimal solutions to be obtained much faster than traditional GA's. The distributed implementation of the GA is developed and tested for a real-world transportation network. Following the development of the algorithm, several computational experiments are performed to assess the impact of varying several of the GA's control parameters on the quality of the solution and the runtime of the algorithm. Results from the study demonstrate that parallel computing has the potential to result in significant reductions in the execution time required to solve the dynamic traffic routing problem. The study also shows that for the dynamic traffic routing problem considered in this study, varying the probability of the mutation operator appears to have a more significant impact on the solution quality than varying that of the crossover operator, and that increasing the number of generations is more beneficial than increasing the population size.