Offline Bayesian Change Point Detection for Carrier-Phase in GNSS-Reflectometry

Change detection techniques are essential for identifying and quantifying variations in data over time, with applications in remote sensing, signal processing, and environmental monitoring. The ability of GNSS-Reflectometry (GNSS-R), a bistatic radar technique that utilizes navigation satellite signals as signals of opportunity, to analyze surface characteristics has recently been explored by studying the coherence of reflected GNSS signal phases. However, detecting changes in the carrier phase of multiple reflections GNSS signals remains unexplored in the literature, as conventional change detection methods are designed for linear data, whereas phase data is inherently circular. In this work, we adapt an offline Bayesian Change-Point Detection approach, assuming the phase noise follows a Von Mises distribution, to detect changes in the reflected GNSS signal by estimating change points through minimizing a contrast function. The method is evaluated using both synthetic and real GNSS-R measurements, demonstrating superior performance compared to a conventional spectral change detection technique. Results highlight the method's accuracy in detecting surface variations in real GNSS-R observations.