Review and prospect of floating car data research in transportation

With the advancement of intelligent transportation systems, floating car data (FCD), as a crucial source of transportation information, has garnered increasing attention for its applications and development directions within the context of massive traffic data. This study conducts an in-depth literature review analysis of FCD in the transportation field based on the Web of Science (WOS) database from 2000 to 2023, employing bibliometric methods and knowledge graph technologies. The current research status was visually analyzed through the literature distribution by year, research regions and institutions, research hotspots, and literature clustering using the bibliometric tool CiteSpace. Three major research topics were identified based on the literature clustering analysis. A systematic review of key literature was conducted to address research challenges related to floating car sampling proportions and frequencies, and future research challenges and opportunities were proposed. The results show an overall parabolic increase in publication volume, with research hotspots mainly focusing on mountainous cities, cluster analysis, machine learning, and deep learning. The three major research clusters include traffic flow state, traffic safety, and route planning. The optimal investment proportion for floating cars is determined to be 3%–8%, and the sampling frequency significantly affects the accuracy of vehicle speed and heading angle information, while having a weaker impact on positional parameters. With the trend of large-scale Internet-connected vehicle deployment in the future, a massive amount of FCD will be generated, prompting in-depth research on the fusion of heterogeneous data sources, including FCD. Future research could focus on leveraging transformer and graph neural networks to explore spatiotemporal features of data, developing lightweight real-time FCD processing algorithms, and constructing multimodal refined models tailored to specific traffic scenarios. • Bibliometric analysis of floating car data research in the transportation field was performed using CiteSpace technology. • Literature publication, keyword burst detection, co-occurrence, and cluster analysis were conducted. • High-frequency floating car data can be used to model continuous operating speed prediction for heavy trucks. • Transformer and Graph Neural Networks may capture the spatio-temporal patterns of floating car data.