Research And Design On Dynamic Load Balancing Algorithm Of Distributed Traffic Simulation

It is a favorable way to improve the efficiency of urban transport and mobility by using a distributed traffic system to simulate realistic traffic conditions. The balance of task is related to the simulation speed of the simulation systems in distributed simulation system. Therefore, it becomes an urgent problem that how to balance the load of the point in distributed traffic simulation platform. This paper is based on the distributed traffic simulation platform and game theory. then we propose a dynamic load balancing algorithm based on game theory. The simulation experiments and theoretical analysis validate the superiority of the algorithm. The main work of this paper includes the following aspects:(1)A model of traffic network based on distributed traffic simulation was proposed. We have designed a mixed model of dynamic load balancing based on the distributed traffic simulation which is independent research and development. This model uses both CS and P2 P. There is a CS model between the master and the terminals and a P2 P model between the terminals. Simultaneously, in order to improve the speed of simulation and reduce the costs of data collection, this paper simplifies the road as a point with weights and the cross as an edge with weights too. Use this model to simulate the real traffic network can simplifies the network model and improve the speed and efficiency of simulation without losing the authenticity of the simulation.(2)A dynamic load balancing algorithm based on game theory was proposed. This paper has referenced the idea of game theory. In the algorithm, road will be actively to migrate. This paper has set up a cost function to the road and proved that Nash equilibrium exists with the function by theoretical derivation. Use an iterative growth algorithm to initial segment the network. The road will calculate its cost by the cost function and decide to migrate or not. The road will also choose the terminal to migrate by the cost function until the all network is balancing. And the road still can do this while the system is running, which can make the unbalancing system to be balancing again.(3)We designed and completed an experiment to compared with the traditional recursive bisection algorithm by using the distributed traffic simulation which is independent research and development as a platform. The main contents of the experiment include: the total simulation time, the variance simulation time, the total communication time, the variance communication time. And the outstanding performance of our model and algorithm were proved.The theoretical and the simulation experimental results that the dynamic load balancing algorithm based on game theory can be used to solve the load-balancing problem of distributed microscopic traffic simulator. On the other hand, the model and the algorithm proposed above can be extended to other distributed simulation system.