On-line Model Learning for Adaptive GNSS Ionospheric Scintillation Estimation and Mitigation
Jordi Vilà-Valls, Centre Tecnològic de Telecomunicacions de Catalunya (CTTC/CERCA), Spain; James T. Curran, European Space Agency (ESA), The Netherlands; Pau Closas, Northeastern University, USA; Carles Fernández-Prades, Javier Arribas, CTTC/CERCA, Spain
Global Navigation Satellite Systems (GNSS) is the technology of choice for most position-related applications. GNSS were initially designed to operate under clear-sky nominal conditions, thus, position accuracy, availability and reliability are compromised under real-life harsh propagation conditions. Among them, ionospheric scintillation is one of the major threats, particularly affecting modern high-precision GNSS receivers which employ carrier phase measurement, using real-time-kinematic (RTK) and precise-point-positioning (PPP) solutions. In recent years there is an increasing interest in ionospheric scintillation estimation and mitigation, because it jeopardizes the use of GNSS in safety-critical applications and critical infrastructure, mainly in equatorial and polar regions.
Under ionospheric scintillation conditions, carrier synchronization is the most challenging receiver stage. Traditional phase-locked loop (PLL) and Kalman filter (KF)-based architectures  fail to mitigate such disturbance, because they track the complete signal phase, thus being incapable of distinguishing between the line-of-sight (LOS) phase of interest and scintillation-induced phase variations (estimation vs mitigation trade-off) . Therefore, it is important to develop improved signal processing techniques for effective carrier tracking in modern GNSS receivers. In previous contributions [2-4], we proposed to model the ionospheric scintillation amplitude and phase components as scalar autoregressive (AR) processes. This enables the use of an augmented state-space formulation for the carrier tracking problem, including both LOS and scintillation amplitude/phase contributions. Using both synthetic scintillation data generated with the Cornell Scintillation Model (CSM) and real scintillation data, this has been shown to be a very promising approach, overcoming the main limitations of standard techniques and providing good scintillation mitigation capabilities.
To enable real-life applicability of these solutions, two key points on the filter design should be addressed: i) on-line AR model order selection, and ii) estimation of the AR model parameters in a recursive manner within the filter. In [2-4], the AR model order was estimated off-line, based on historical data, and then used onwards in the tracking process. Similarly, the AR model parameters were computed off-line by fitting the model to synthetic [2-4] and real data . This was an approach that produced clearly improved performance, however, to account for possibly time-varying propagation conditions the filter should adapt these parameters.
In this article we further elaborate on our previous scintillation mitigation methods, which considered AR scintillation approximations, with the following main contributions: i) we propose an on-line adaptive model learning method to iteratively estimate the AR parameters within an extended KF tracking architecture; and ii) we show the carrier tracking and on-line model estimation capabilities of the new approach using real ionospheric scintillation data, recorded over Hanoi in April, 2015.
 J. Vilà-Valls, P. Closas, M. Navarro, and C. Fernàndez-Prades, “Are PLLs dead? A tutorial on Kalman filter-based techniques for digital carrier synchronization,” IEEE Aerospace & Electronic Systems Magazine, vol. 32, no. 7, pp. 28-45, July 2017.
 J. Vilà-Valls, P. Closas, C. Fernàndez-Prades, J. A. López-Salcedo, and G. Seco-Granados, “Adaptive GNSS carrier tracking under ionospheric scintillation: estimation vs mitigation,” IEEE Comm. Letters, vol. 19, no. 6, pp. 961–964, June 2015.
 J. Vilà-Valls, P. Closas, and C. Fernández-Prades, “Advanced KF-based methods for GNSS carrier tracking and ionospheric scintillation mitigation,” in Proc. of the IEEE Aerospace Conference, March 2015.
 J. Vilà-Valls, P. Closas, and J. T. Curran, “Performance analysis of multi-frequency GNSS carrier tracking for strong ionospheric scintillation mitigation,” in Proc. of the European Signal Processing Conference (EUSIPCO'17), August 2017.
 J. Vilà-Valls, P. Closas and J. T. Curran, “Multi-frequency GNSS Robust Carrier Tracking for Ionospheric Scintillation Mitigation”, Journal of Space Weather and Space Climate, vol. 7, A26, October 2017.