Application of state-space frequency estimation to a 24-GHz FMCW tank level gauging system

Radar is a well established tank level gauging technique. Today, a shift towards more demanding measurement applications can be observed. As a result, novel radar signal processing algorithms emerge. One of the most challenging tasks for tank level gauging systems is separation of closely spaced targets, especially with the given bandwidth limitations set by both legal means and technical feasibility. Superresolution methods are expected to overcome the inherent resolution limits of Fourier transform based algorithms. However, they are fairly complex, making adaptation to real world radar signals a tedious and complicated task. Furthermore stability problems occur, mainly caused by the necessary a priori model-order estimation. This paper now demonstrates a robust implementation in a commercial 24 GHz radar tank level gauging system, where an adaptive model-order estimation algorithm is used. The measurements in an industrial tank show a threefold increase of the resolution while maintaining stability. Finally a special set of measurements shows that the selected state-space approach is especially well suited for the resolution of small peaks adjacent to larger echoes.