Benefits of Adaptive Kalman Filter-Based Single Point Positioning in Dense Urban Environments

Abstract

This paper reveals that as compared to the effects achieved using a normal extended Kalman filter (EKF), an adaptive EKF improves the performance of Global Navigation Satellite System (GNSS) positioning and its integrity information in dense urban canyons. An important and challenging problem related to such canyons is the negative impact of non-line-of-sight (NLOS). Large buildings shade line-of-sight (LOS) signals from GNSS satellites and create reflecting or diffracting NLOS signals; eventually, GNSS receivers track only the NLOS signals. In such severe environments, significant position outliers arise because of measurement from NLOS-only tracking. To reduce the negative impact of NLOS, an adaptive EKF is implemented for GNSS single point positioning. No sensor aiding or coupling is employed, despite their popularity in recent times. The test results show that the adaptive EKF drastically improves positioning accuracy and precision in dense urban areas. The percentage of horizontal position errors within 20 m is 95.3% for the adaptive EKF and 42.7% for the normal EKF. The EKFs in this study only differ in the adaptive estimation of the measurement noise covariance matrix. A detailed analysis confirms that the adaptive covariance matrix fairly matches the actual measurement error, leading to great improvement in positioning accuracy and precision. Further, NLOS-only tracking causes remarkable random measurement errors, frequently exceeding 50 m. It seems very challenging for the normal EKF to tackle such errors. Moreover, the adaptive covariance matrix contributes to integrity information. The horizontal protection level (HPL) computed by the covariance matrix of the state vector is degraded for the normal EKF, as 87.0% of positions exceed its HPL, while that of the adaptive EKF is maintained, as only 11.4% of the positions exceed its HPL. Thus, the adaptive EKF is effective for improving GNSS performance in dense urban canyons.