Study On The Floating Car Map-matching Algorithm Based On Storm

Floating car technology is widely used in intelligent transportation systems(ITS) as one of the technical means for acquiring road traffic information. A large number of buses & taxis, which equipped with GPS positioning system, continuously sent its vehicle information(such as time, coordinates, speed and direction angle, etc.) to a data center. After analysing and processing the real-time GPS data, we can get the real-time traffic status information in one area. But due to the low precision of GPS data, when combined with digital map, it may lead to the phenomenon that the vehicle position deviates from a road, so we need to use a map-matching technology to correct the error. The map-matching technology associates vehicle GPS information from GPS devices with the urban road network topology by the algorithm model, and projects the GPS observation point to the real road. For the large, poor accurate floating car data, how to effectively guarantee the matching efficiency and accuracy? Map matching algorithm is the key for subsequent other studies(e.g., improving the quality of road traffic state analysis).To solve above problems, this paper proposes a map matching algorithm based on spatial-temporal analysis method. The algorithm could be divided into two parts, including the digital map data preprocessing and spatial-temporal analysis of the candidate point. In map data preprocessing part, an grid index was built for Chongqing road network based on Arc GIS platform, which greatly improved the efficiency of searching for roads surrounding the GPS observation point. Spatial-temporal analysis section is the core of the algorithm. The algorithm comprehensively considers the space geometry, the road network topology information and time factor influence on GPS candidate selection, which improve the matching accuracy.Considerating the amount of data confronted by the floating car data center is increasing rapidly, map-matching algorithm on a single machine is more and more difficult to meet the demand of data processing. This paper explored the distributed real-time computing technology’s application in traffic area, and studied the most popular real-time computing technology-- Storm. At last, the paper successfully migrated the proposed map matching algorithm to Storm platform.Finally, using the GPS data bus in Chongqing and electronic map data on the Arc GIS platform, the paper verified the performance of the algorithm(e.g., accuracy, efficiency) respectively in stand-alone point and Storm cluster environment. The results show that the spatial-temporal analysis algorithm significantly improves the accuracy and efficiency compared with other algorithms. The algorithm based on Storm platform can take full advantage of machine’s multi-core characteristics, as well as facilitate horizontal expansion of cluster scale, which promote the efficiency linearly on the basis of the original algorithm.