On Micro-Doppler Period Estimation

A novel approach dedicated to estimation of period in micro-Doppler radar signatures represented in the form of time-frequency distribution is suggested. The approach is based on the exploiting short-time Fourier transform performed with window sliding along the micro-Doppler spectrogram. No a priori information about radar object under classification is required for estimation of period. Modeled data of human walking is used both for performance evaluation of suggested and common approaches. Real-life micro-Doppler radar measurements of a walking human are collected for indoor environment and used for experimental examination of suggested approach. Experimental radar measurements of toy helicopter rotating blades are executed and parameters of helicopter micro-Doppler radar signatures are studied. The benefit of suggested approach comparing to common approaches is demonstrated using the bias and mean squared errors of micro-Doppler period estimates computed in additive noise environment.