Generalized Robust Adaptive Control Algorithm for GNSS Receivers

Abstract

Synchronization is the core of global navigation satellite system (GNSS) receivers. Successful synchronization with the incoming GNSS signals leads to estimating the pseudorange and pseudorange rate and calculating a reliable position, velocity, and time (PVT) solution. However, a fixed configuration of the synchronization stage challenges its capability in time-varying scenarios, where noise and dynamic variations may lead to loss of lock and hinder the achievement of a PVT solution.

This thesis introduces an adaptive control algorithm to strengthen the GNSS receiver’s synchronization robustness. The adaptive control algorithm updates the response time of a plurality of scalar tracking loops (STLs). The update consists of a response-time-dependent weighted difference between estimated noise and dynamic information from the STL.

A particular implementation of the proposed algorithm is the loop-bandwidth control algorithm (LBCA), which adapts the loop bandwidth of a single STL. The LBCA performs a weighted difference between the dynamic and noise statistics derived from the discriminator’s output to update the loop bandwidth. These weights are determined by a linear combination of Sigmoid functions, which are dependent on the current loop bandwidth. To ensure practical, low-complexity, and effective implementation of the LBCA, careful considerations such as selecting the appropriate statistics, piecewise-linearizing the Sigmoid functions, and incorporating a Schmitt trigger are considered.

The LBCA is integrated into advanced tracking schemes such as the direct state Kalman filter (DSKF) and the lookup table (LUT)-DSKF. The former method uses the LBCA to update the process noise variance, whereas the latter directly relates the loop bandwidth with the steady-state Kalman gains. The LUT-DSKF stands out for its reduced complexity due to the derivation of the steady-state Kalman gains.

This research performs a comparative analysis between the LBCA-based tracking schemes and other adaptive techniques like the fast adaptive bandwidth (FAB), fuzzy logic (FL), and adaptive DSKF. These adaptive tracking techniques are implemented in an open software interface GNSS hardware receiver. The receiver’s carrier and code system performance for each implementation is evaluated in simulated scenarios with different dynamics and noise levels. Results demonstrate the successful implementation of the LBCA in the DSKF. Additionally, the LBCA-based STL and LBCA-based LUT-DSKF present better dynamic system performance than the LBCA-based DSKF, which is limited by its slow convergence time. Furthermore, the superiority of the LBCA-based tracking schemes over the other adaptive tracking techniques is observed while maintaining low complexity. The LBCA demonstrates robust tracking sensitivity in noisy static scenarios and resilience to dynamics in high-dynamic scenarios, highlighting its advantages over fixed tracking configurations.

Furthermore, this work applied the adaptive control algorithm in advanced tracking architectures, mainly in long-coherent integration and vector tracking architectures. An extension of the LBCA is presented: the normalized bandwidth control algorithm (NBCA). The NBCA updates the loop bandwidth and the integration time of the tracking loop, helping the transition to long-coherent integration architectures. However, the dynamics experienced in carrier tracking limit the adaptation range of the carrier’s integration time. To fully exploit the NBCA’s performance, a combination of long-coherent integration and ultra-tight integration architectures is shown. The NBCA improves the synergy between tracking channels and the navigation engine by adapting the response time of the tracking channels, weighting the measurement covariance matrix of the navigation engine, and deciding between PVT-based estimated Doppler and internal Doppler of the tracking channels. The adaptive ultra-tight integration architecture further improves the system’s performance in harsh scenarios. Also, the rapid relock and hot start capabilities of this architecture significantly reduce the power consumption of a GNSS receiver by using the acquisition engine minimally.

The generalization of the adaptive control algorithm is demonstrated by its successful implementation in the interference mitigation stage. Specifically, the LBCA was integrated into adaptive notch filters (ANFs) to adjust the response time of the frequency-locked loop (FLL), which estimates the frequency of the interference. The results showed improved performance compared to fixed configuration settings. Furthermore, further research expanded upon this concept by developing a more sophisticated algorithm, the multi-parameter adaptive notch filter (MPANF). The MPANF adapts the notch width, the notch depth, and loop bandwidth of the FLL, further enhancing the ANF’s adaptability and effectiveness in mitigating interference.

The adaptive control algorithm is a parametric-based system. Its main limitation resides in the correct tuning of the weighting functions, the proper selection of the noise and dynamic estimates, and the interdependency between tracking loops. Therefore, a reinforcement learning framework is proposed as a potential solution to optimize these parameters.

The potential of the adaptive control algorithm is evident, and its applicability extends beyond the domain of GNSS, finding relevance in control systems where measurements exhibit time-varying noise and dynamic levels.