Multi-spectral echo signal processing for improved detection and classification of radar targets

A major challenge in radar based remote sensing and imaging is to identify and to detect radar targets, and also to accurately determine their locations and sizes. This especially applies in the case of multiple, spatially distributed radar targets, as for example in radar imaging, automotive radars, and others. Previously, we have proposed a concept for multi-spectral analysis and processing of echo signals for radar level measurement of bulk solids in silos using a spatially fixed antenna beam. This approach has now also been utilized for scanning radar applications. The basic technique is to filter the radar echo signals in multiple frequency sub-bands and to incoherently combine the filtered signals. Furthermore, the variance of envelope signals is analyzed in order to allow for a differentiation between echoes from distributed, randomly arranged scatterers and from spatially isolated single scatterers. The mean over the standard deviation of the envelope signals obtained from the different sub-bands is suggested to be used as an amplitude-invariant parameter for the classification of radar targets. Results of an experimental evaluation of the concept using a mechanically scanning 75 to 80 GHz Frequency Modulated Continuous Wave (FMCW) radar system are presented. It will be shown that the proposed technique enables to largely suppress the echo signal fluctuations, which are given in scenarios of spatially distributed radar targets, and also to distinguish between different kinds of radar targets.