A 24-GHz tank level gauging system with state-space frequency estimation and a novel adaptive model order selection algorithm

For the evaluation of signals in frequency modulated continuous wave radar systems for object distance measurement tasks, highly sophisticated state-space frequency estimation algorithms pose an alternative to the traditionally used fast Fourier transformation. Current implementations of this approach suffer from instability problems when applied to real radar measurements. In this paper a method is presented to overcome the present stability problems by proposing a novel model order estimation procedure. With the use of practical measurements in a level gauging system an improvement of 2-3 over the resolution of the FFT-based methods is shown, and stability of the model order selection algorithm is proved.