This paper aims to improve the marginalized likelihood ratio test (MLRT) algorithm for detecting GNSS spoofing attacks in integrated GNSS/INS systems, thereby ensuring secure and reliable navigation for aerospace applications. MLRT is used for spoofing detection in GNSS/INS navigation through consistency checks between satellite and inertial data. However, standard MLRTs often assume static and known noise statistics. In practice, this assumption degrades the detection performance against complex, slow-growing spoofing attacks.
To address this issue, a self-tuning MLRT is proposed, where the Inverse-Wishart distribution is used as a conjugate prior for modelling uncertainties in the observation and process noise covariances.
Monte Carlo simulations of a loosely coupled GNSS/INS architecture implemented on an aerial vehicle demonstrate that the proposed detector outperforms fixed-parameter methods in terms of detection latency, offering superior sensitivity to slow-growing spoofing.
To the best of the authors’ knowledge, the originality of this paper lies in overcoming the limitations of static noise assumptions by incorporating the Inverse-Wishart distribution into the MLRT approach, thereby enhancing the resilience of navigation systems.
