For the purpose of better accuracy and higher coverage rate of traffic information provided by a traffic estimation system, a large quantity of probe vehicle data from different sources is indispensable. It is important to deal with the scalability problem efficiently as the growth of probe vehicle data. In this paper, a distributed stream processing architecture is deployed to tackle the scalability problem of real-time multi-source traffic sensing and analyzing. We show the ideal characteristics and advantages of distributed stream processing that are applicable to an intelligent transportation system. The system data flow for estimating urban traffic on a distributed stream processing platform is then presented. Finally, the implementation results show that the distributed stream processing can be applied to real-time multi-source traffic estimation system effectively and efficiently.
展开▼