Traffic Flow Modeling And Implementation Of The Emulation System

There are many ways to emulate the traffic flows. Not only can we analyze and emulate the historical statistics but also emulate the dynamic traffic flows. The former method is useful for traffic drainage and modeling while the other one aims at traffic prediction. Using the existing model and data settings, we do the emulation work on a computer to predict the situation of related highways. Two types of traffic modeling coexist, which are macroscopically and microcosmic modeling.Nowadays, microcosmic modeling is mainly used for metropolitan traffic system design and evaluation. Or, we can also evaluate the coming traffic facilities and strategies according to the outcome of emulation. In order to reveal the complicated traffic flows, we start from analyzing the traits of some related traffic models. After that, by adopting the microcosmic traffic modeling, we model the traffic flow based on traffic flow theories. Afterwards, we will design and implement a parallel micro traffic flow emulation platform, on which the traffic models runs so that we can observe the virtual traffic flow. And, we will improve the traffic models according to the traits of real traffic flow and the traffic models in using. Finally, we would simulate the traffic flow on real road, and compare the statistics of the emulation system with that from reality, and offer a thorough optimization for the emulation platform and traffic models as well. By this way, we can achieve a relatively complete micro traffic flow emulation system. It has the following advantages. First, it can show the running traffic flow in time. Secondly, it can provide the statistics in time. By the way, in our experiments, the emulation statistics are close to those broadly used mathematics models.This traffic flow emulation system, implemented with C++, runs on a parallel computer which has distributed memory system and Center OS installed on it. We solve the problem of data transportation and implement parallel computation with MPI. We also find that, by dividing the road net properly, the system efficiency can get close to 50%. The emulation system can give a mimic traffic flow similar to real ones by the adoption of the performance parameters of real cars. Unfortunately, we find it hard to improve the system efficiency to an ideal extent because of the random traffic activities. Therefore, we have to leave the tasks distribution to the future stage, on which we will add the traffic signals and the structure of real road net as well. All in all, although this traffic flow emulation system still has not met the commercial standards, it can be utilitarian in large scale and various complicated emulation tasks.