Realtime vehicle routes optimization by cloud computing in the principle of TCP/IP

With the advancement of the urbanization process, traditional centralized dynamic route guidance systems become incapable of handling fast growing traffic large data, due to the computational complexity. In another aspect, the realtime vehicle routing services from customers, such as information pushing and travel time forecasting, are more demanding. Aimed at easing the traffic congestion pressure and improving the quality of transportation services, this paper presents the idea of decentralized vehicle routing service system (DVRSS), in the principle of TCP/IP approaches. With the cloud computing architecture, our system adopts the MapReduce method in Hadoop to fulfill the task assignments in parallel, uses ZooKeeper technology to establish the coordination system between subtask processors, and applies Kalman filter algorithm to conduct traffic flow prediction in short term. In this way, DVRSS is able to provide vehicle routing services in realtime, achieve more effective communications, and provide effective route recommendations, accurate travel time prediction, and versatile push notification services.