Research On Large Scale Parallel Computing Method Of Urban Traffic Flow

Computer simulation technology is the core of the modern high technology,many fields has become a good tools to analysis, evaluation and operation. For theresearch of urban traffic problems, simulation technology is also importantexperimental methods and tools of traffic management control and decision-makingthat can accurately reflect the real traffic activity. Before the "real" test or expensiveroad-building program, must study the impact on traffic congestion of change the roadnetwork or traffic control mechanism on the computer. There by it can be reducewaste of money. Traffic simulation is the use of scientific theory to solve thetransportation planning, management and control. With the social development ofurban traffic network is growing larger and larger, the requirements of the trafficsimulation technology more and more high. Parallel simulation of traffic flow is thefurther development of the traffic simulation technology. In recent years, the trafficparallel computing is a popular method to solve the urban traffic flow simulationproblem.Mainly work of this paper is designing and complete based on Linux distributedparallel computing platforms. Proposed parallel simulation algorithm of Local trafficflow-based on the MPI message passing library. Design a master-slave parallelsimulation algorithm with the C++language. Consider the characteristics of the roadnetwork, using domain decomposition algorithms for task allocation of the simulation.Last part summarized the research and analysis of research results. To see from theresults of parallel simulation, this parallel system is can be used, reliable. And thisalgorithm can improve simulation speed.