A Novel Two-Stage 2D-CFAR Detector for Automotive 77-GHz DDMA-MIMO Radar

For automotive multiple-input multiple-output (MIMO) radars, constant false alarm rate (CFAR) techniques are typically used to detect unknown targets. To minimize the omission rate, this paper proposes a novel 2D-CFAR detector utilizing a two-stage processing approach for Doppler division multiple access (DDMA) MIMO radars. In the first stage, a classic 2D-CFAR detector with large threshold and detection windows is employed to extract dominant targets. In the second stage, for cells on range-Doppler maps (RDMs) excluding those identified as targets in the first stage, smaller threshold and detection windows with a different threshold in the 2D-CFAR detector are used to retrieve weaker targets. By using an automotive $77-\mathrm{GHz}$ DDMA-MIMO radar system, experimental results in a parking lot scenario showcase denser, more precise, and more complete point clouds compared to the common 2D-CFAR detector, demonstrating the superiority of our proposed technique.