[Paper Review] CFAR-Based Interference Mitigation for FMCW Automotive Radar Systems

Jianping Wang|arXiv (Cornell University)|Jan 4, 2021

Radar Systems and Signal Processing23 references4 citations

TL;DR

This paper proposes two CFAR-based interference mitigation techniques—CFAR-Z and CFAR-AC—for FMCW automotive radar systems. By detecting interference as short chirp-like components in the time-frequency domain using a 1D CFAR detector and applying zeroing or amplitude correction, the methods effectively suppress interferences while preserving useful target signals, achieving superior signal-to-interference-plus-noise ratio (SINR) and correlation coefficient compared to existing methods, with low computational complexity suitable for real-time implementation.

ABSTRACT

In this paper, constant false alarm rate (CFAR) detector-based approaches are proposed for interference mitigation of Frequency modulated continuous wave (FMCW) radars. The proposed methods exploit the fact that after dechirping and low-pass filtering operations the targets' beat signals of FMCW radars are composed of exponential sinusoidal components while interferences exhibit short chirp waves within a sweep. The spectra of interferences in the time-frequency ($t$-$f$) domain are detected by employing a 1-D CFAR detector along each frequency bin and then the detected map is dilated as a mask for interference suppression. They are applicable to the scenarios in the presence of multiple interferences. Compared to the existing methods, the proposed methods reduce the power loss of useful signals and are very computationally efficient. Their interference mitigation performances are demonstrated through both numerical simulations and experimental results.

Motivation & Objective

• To address the growing challenge of interference in dense automotive radar environments where multiple FMCW radars operate simultaneously.
• To develop a computationally efficient, real-time interference mitigation method that preserves weak target signals degraded by strong interferences.
• To overcome limitations of existing methods such as signal power loss, reliance on prior signal models, or complex system redesign.
• To exploit the distinct spectral characteristics of interference (short chirps) versus target beat signals (exponential sinusoids) in the time-frequency domain for effective detection and suppression.

Proposed method

• A 1D CFAR detector is applied along each frequency bin in the time-frequency (t-f) domain to detect interference components, which appear as short chirp-like signals due to their transient nature.
• The detected interference map is dilated into a binary mask to identify interference-affected regions in the t-f plane.
• For CFAR-Z, the identified interference regions are zeroed out in the t-f spectrum, followed by inverse short-time Fourier transform (ISTFT) to reconstruct the beat signal.
• For CFAR-AC, amplitude correction is applied to the interference-affected regions using a gain factor derived from neighboring non-interference regions, preserving signal energy.
• The reconstructed beat signals are used to generate range profiles, and target detection performance is evaluated using a CFAR detector.
• Zero-padding is applied before STFT and removed after ISTFT to minimize edge effects from windowing.

Experimental results

Research questions

• RQ1 Can a CFAR-based detection approach effectively distinguish interference components from target beat signals in the time-frequency domain of FMCW radar signals?
• RQ2 How does CFAR-Z compare to CFAR-AC in preserving useful signal power while suppressing interference?
• RQ3 What is the impact of interference mitigation on target detection performance, particularly for weak or closely spaced targets?
• RQ4 Can the proposed methods achieve real-time processing with low computational overhead compared to existing signal processing or deep learning-based approaches?

Key findings

• The CFAR-AC and CFAR-Z methods achieve the highest correlation coefficient and signal-to-interference-plus-noise ratio (SINR) among all tested methods, outperforming wavelet denoising (WD) and adaptive noise canceller (ANC).
• CFAR-AC and ANC methods produce the highest peak amplitudes in the range profiles at target ranges, closest to the reference signal, indicating better preservation of useful signal power.
• The CFAR-Z method results in lower peak amplitudes than CFAR-AC, indicating partial loss of useful signal energy due to aggressive zeroing.
• The CFAR-Z and CFAR-AC methods successfully detect a fourth stationary target at 22.5 m that is missed by WD and ANC methods, demonstrating improved detection probability.
• All four methods (WD, ANC, CFAR-Z, CFAR-AC) improve target detection compared to the raw signal, with CFAR-Z and CFAR-AC enabling detection of previously masked weak targets.
• The proposed methods are computationally efficient and suitable for real-time implementation in automotive radar systems, unlike methods requiring complex signal redesign or high sampling rates.