Person Identification from Millimeter-Wave Radar micro-Doppler Signature

Radio frequency indoor sensing technologies are becoming a widely studied topic in recent years for security and monitoring applications. In this thesis the problem of person identification from the properties of the reflected signal is addressed. A novel processing pipeline is presented, including a denoising phase, a clustering procedure and a classification based on deep learning algorithms. The evaluation is conducted on a public dataset and on data measured with a mm-wave radar.