Sensor Data Fusion for Seamless Navigation using Wi-Fi Signal Strengths and GNSS Pseudoranges

摘要

This paper proposes an information fusion approach for seamless indoor and outdoor positioning in urban scenarios. In these scenarios the global navigation satellite systems (GNSS) easily reach their limitations, whereas Wi-Fi fingerprinting positioning systems profit from signal degradation such as shadowing and reflection. The solution presented here is based on a Bayesian framework, fusing data from a Wi-Fi fingerprinting algorithm with pseudoranges of a GNSS receiver. A particle filter is used to combine the Wi-Fi database correlation results computed on a discrete fingerprint grid with the pseudorange measurements. Further, the additional estimation of the GNSS user clock offset, allows this approach to be seen as a Wi-Fi aided position, velocity and time determination algorithm (PVT) of a GNSS receiver. The presented algorithm has the ability to improve the Wi-Fi fingerprinting algorithm with less than four pseudoranges available. A filter to solve the typical Wi-Fi fingerprint positioning ambiguities has been developed. This algorithm achieves a higher robustness and accuracy, compared to standalone Wi-Fi positioning or GNSS, especially in urban canyon scenarios.