Parametric estimation of distances with a 122 GHz FMCW radar system

Industrial applications for frequency modulated continuous wave (FMCW) radar systems often require the separation of close targets, which is limited by the available distance resolution of the radar. In general, for the evaluation of FMCW radar signals a fast Fourier transformation (FFT) based method is used. However, the FFT inherently limits the available distance resolution in dependence of the bandwidth. To overcome this limitation, super resolution algorithms or parametric methods can be used. A 122 GHz FMCW radar with a bandwidth of 1 GHz is used to take measurements of two targets, below the FFT limited distance resolution (150 mm) of the FMCW radar. It is shown by using an expectation maximization based algorithm that it is still possible to distinguish between both targets over the whole measurement distance. The maximum distance error remains in the millimeter range as long as the targets are separated by at least 50 mm, which equals an improved distance resolution of a factor 3 compared to the FFT. Furthermore, the influence of model order errors on the expectation maximization algorithm are analyzed.