Stationary clutter- and linear-trend suppression in impulse-radar-based respiratory motion detection

Radar-based respiratory motion detection, implemented in the case of e.g. detection of buried victims in post-disaster scenarios, requires removal of radar returns from motionless objects. However, time-domain radar, which is considered herein, often shows linear amplitude instability in the time base. This can lower the probability of detection. This paper investigates the performance of three stationary-clutter subtraction methods: range profile subtraction (RPS), mean subtraction (MS) and linear-trend subtraction (LTS) method. The performance evaluation is performed on measured data containing respiratory-motion response and linear amplitude instability. It has been shown that the RPS method should be avoided since it increases the noise power and acts as a differentiator, thereby resulting in frequency-dependent noise floor. The MS and LTS method show identical performance in the absence of any amplitude instability. However, in the presence of it, the LTS method performs better and is therefore the preferred method for use in impulse-radar-based respiratory motion detection.