Vehicular Positioning Using mmWave TDOA with a Dynamically Tuned Covariance Matrix

5G/6G small cells have the potential to enable sub- meter positioning accuracy in urban canyons and downtown scenarios, where global navigation satellite systems (GNSS) suffer the most. In order offer a robust 5G/6G time-based trilateration position solution, extended Kalman filter (EKF) is usually utilized to estimate the position. One of the main drawbacks of EKF lies in its linearization of state dynamics and processes, which would inevitably induce linearization errors. Such errors would propagate through the filter, which will eventually lead to positioning errors. In this paper, the analysis of the fundamental causes of such errors is undergone. Additionally, we propose to dynamically tune the Kalman filter’s measurement covariance matrix to automatically exclude base-stations (BSs) that induce high linearization errors, hence, mitigating the limitations of the EKF. The performance of the proposed method was tested against the traditional implementation of the EKF using realistic 5G/6G signal propagation data, obtained from Siradel’s S_5GChannel simulator. Our simulations include two realistic trajectories in downtown Toronto that are 1.63km and 2.69km long, respectively. The results show that the proposed method outperforms traditional the EKF implementation in both trajectories, as the 2D RMS/maximum positioning errors were reduced by 71.76%/84.11% and 51.18%/35.9% for the first and second trajectories respectively.