Toward Accurate Inter-vehicle Distances based on GNSS Pseudorange Measurements in a non-Gaussian Generalized Error Distribution Environment

Intervehicle accurate positioning information is a cardinal prerequisite for proper and efficient management for many vehicular applications, such as collision avoidance, navigation, and intelligent transportation systems (ITSs). To improve the positioning accuracy, collaborative vehicular localization techniques based on fused data from different sources have been proposed in vehicular ad hoc networks (VANETs). Although the existing positioning techniques achieve high precision, they heavily rely on large-scale infrastructures and/or on the support of specific hardware devices, some of which are not generally accessible or are expensive, so they are impractical in many scenarios. In this article, a novel collaborative vehicular localization approach termed non-Gaussian weighted least-squares ( ${n}$ GWLS) is proposed to estimate the intervehicle distances. In this approach, the global navigation satellite system (GNSS) pseudorange double-difference (PDD) measurements are adopted, whereas the position dilution of precision (PDOP) and the number of common satellites of two given GNSS receivers are used to evaluate the quality of the PDD. To achieve reliable pseudorange measurements, the uncommon noise specific to a GNSS receiver and satellite is utilized as non-Gaussian distributed, where the generalized error distribution (GED) is adopted as an approximation to non-Gaussian densities. Moreover, the weighted least-squares technique is exploited, which considers the signal quality of each pseudorange measurement. The comparison study based on real-world experiments and statistical analysis demonstrates that the proposed method can effectively obtain an accurate intervehicle distance estimation.