The Doppler signature of a man walking in a forested area analysed at L-band is presented here. The aim is twofold: to assess the best time-frequency distribution to characterise the activity; to highlight the similarity of the simulated data to the measured ones to validate the simulation tool. Indeed, the Doppler-Time (DT) signal variation represents the main characteristic of Artificial Neural Networks (ANNs) for classification. The more accurately the DT characterises the activity, the higher the machine's accuracy in classifying it. Besides, in the training data frame, reliable simulated models may supply the amount of data needed by ANN applications. Thus, a short-time Fourier transform (STFT), a reassigned spectrogram (RE-Spect), and a pseudo-Wigner-Ville distribution have been applied to the measured and simulated data. The measurements have been performed using a bistatic radar working at 1 GHz. Then, the measurement setup has been replicated in simulation, and 3-D human bodies walking in free space have been computed using physical optics. The results show that the STFT is the most suitable time-frequency method for recognising and classifying the walk. Moreover, the simulated data are in agreement with the measured data, regardless of the chosen Cohen's technique.This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.