Radar Signal Processing for Jointly Estimating Tracks and Micro-Doppler Signatures

The aim of the radar systems is to collect information about their surroundings. In many scenarios besides static targets there are numerous moving objects with very different characteristics, such as extent, movement behavior or micro-Doppler spread. It would be most desirable to have algorithms that extract all information on static and moving object automatically, without a system operator. In this paper, we present measurements conducted with a commercially available high-resolution multi-channel linear frequency-modulated continuous-wave radar and algorithms that do not only produce radar images but a description of the scenario on a higher level. After conventional spectrum estimation and thresholding, we present a clustering stage that combines individual detections and generates representations of each target individually. This stage is followed by a Kalman filter based multi-target tracking block. The tracker allows us to follow each target and collect its properties over time. With this method of jointly estimating tracks and characteristics of each individual target in a scenario, inputs for classifiers can be generated. Which, in turn, will be able to generate information that could be used for driver assistance or alarm trigger systems.