The evaluation of walking speed plays an important role in gerontology as it reflects the health and functional status of older people. In this paper, we propose the use of a Doppler radar system with continuous wavelet transform (CWT) analysis for in-home gait characterization. A methodology based on an accurate 3D motion-capture camera system (Vicon) has been developed in order to validate the suitable mother wavelet. The CWT analysis with several mother wavelets has been applied to our experimental Doppler radar signals. The Pearson Correlation coefficient (ρ) has been computed between the gait speed signals obtained from the radar and those obtained from the Vicon system. Our outcomes suggest the use of Daubechies5 and Symlets7 wavelets giving a ρ values of 0.86 and 0.85 respectively with a mean square error value less than 0.05 m/s in comparison with the correct gait speed value.