Extraction of Micro-Doppler Signatures using Automotive Radar Sensors

摘要

Detection of pedestrians in an urban environment is a highly sophisticated task. This paper presents a signal processing technique suitable for frequency-modulated continuous wave (FMCW) radar sensors using chirp sequence modulation, allowing observation of slow-moving objects with high resolution capability in range and velocity. Velocity resolution can be improved through auto-regressive linear prediction (AR-LP) without requiring a long chirp sequence duration. Furthermore, the influence of the frequency deviation used and velocity resolution will be shown in a scenario with two objects. An adaptive velocity resolution improvement will be presented to extract a characteristic micro-Doppler signature (uD-signature) of a slowly moving pedestrian, a car, and an inline-skater. This is verified by measurements at the carrier frequency of 77 GHz.