Multi-sensor heterogeneous data representation for data-driven ITS

In data-driven intelligent transportation systems (ITS), the heterogeneous data collected from various traffic sensors are used in multiple practical applications. To achieve the effective data sharing in different applications, this paper proposes the formal representation of heterogeneous ITS data by calculating their common traffic features and integrating the spatio-temporal-related data. According to different granularity levels of ITS data, the formal representation is outlined as three primitives: cell, compu-band, and computational domain, which are from basic units to their combination. The utilization of cell, compu-band, and computational domain shows that different data granularity levels can provide efficient processing for multi-scale data units. Moreover, the data collected from Sydney Coordinated Adaptive Traffic System (SCATS) loop detectors and Global Positioning System (GPS) probe vehicles are used to construct the ITS computational domain, and the utilization of their integrated data by data fusion for traffic state estimation is investigated to test the availability and effectiveness of the multi-sensor heterogeneous data representation. The experimental results show that the ITS computational domain can implement effective data integration and sharing, and therefore can improve the accuracy and data utilization in various data-driven applications.