Hybrid Algorithm for Small Doppler Shifts Evaluation

Abstract

Improving performance of Doppler shift evaluation could be generally attempted by enhancement of hardware or software. The costly and inflexible nature of the former motivates us to gather as much as possible from the signal processing. The aim of our research has been to develop a real-time algorithm improving precision of Doppler shift evaluation for a standard, low-cost CW (continuous wave) Doppler radar. Emphasis has been put on close-to-zero Doppler shifts (i.e. low velocities), where standard algorithms usually do not provide accurate and reliable results. The proposed algorithm uses simple neural networks in order to provide the approximate estimate of the Doppler shift. According to their results, a proper processing method is chosen for the evaluation of the accurate Doppler shift value.