Contents
DoA Estimation for automotive FMCW Radar
Direction-of-arrival estimation for automotive FMCW MIMO radar, covering the signal model, snapshot extraction, covariance-matrix processing, and the main DoA algorithms: Spatial FFT, Bartlett, Capon/MVDR, MUSIC, and ESPRIT. The methods are organized into three groups: conventional beamforming with the Spatial FFT and Bartlett beamformer, adaptive beamforming with Capon/MVDR, and subspace-based estimation with MUSIC and ESPRIT.
The Jupyter notebooks introduce the FMCW MIMO signal model, snapshot extraction strategies, covariance-matrix estimation and preprocessing, and the assumptions behind each DoA algorithm. Key formulas are complemented by compact Python implementations, numerical examples, and practical engineering observations.
The theoretical foundations and algorithm implementations are validated using real measurements acquired with a Texas Instruments AWR2243 radar. The radar is configured as a two-transmitter, four-receiver TDM MIMO system, forming an eight-element virtual uniform linear array. The validation datasets are provided as part of the workshop:
DoA Estimation for Automotive FMCW Radar
This workshop evaluates angular resolution in controlled broadside and off-boresight scenarios using static corner reflectors. The corresponding workshop and dataset are available here.