Simulation and Analysis of Realistic Human Motion Micro-Doppler Signatures in Through-Wall Radar

Abstract

Obtaining the large statistically independent training datasets for effective classification in through-wall radar studies is always remaining as a challenging task due to the practical and monetary difficulties in the collection of real -time experimental data. Therefore, the development of accurate but computationally less complex simulation models that combine human micro-Doppler data with through-wall propagation effects is instrumental in generating such large training datasets, which can improve the classification accuracy. The proposed research work has presented a simple and accurate simulation methodology which combines human Radar Cross Section (RCS) modeling using primitive based techniques and Finite Difference Time Domain (FDTD) simulation of walls. The human radar scatterings are analyzed in the Line of Sight (LOS) conditions and through-wall scenarios. The simulated spectrograms in the LOS conditions are corroborated by measurement data collected using a monostatic continuous wave Radar. The simulated spectrograms in through-wall conditions shows that the distortions in the Doppler frequencies from different body segments are minimal but the magnitude response of the doppler spectrogram is affected due to attenuation and fading