Automotive Radar Interference Isolation Using STFT and DoA Fusion

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Radar systems in automotive applications face interference issues due to congested spectrum environments, leading to degraded performance and failure, as existing thresholding methods often remove target signals along with interference, reducing signal-to-noise ratio and causing ambiguous sidelobes.

Innovation Solution

Implementing time-frequency domain threshold interference and localization fusion by producing spectrograms using Short-Time Fourier Transform (STFT), determining suppression thresholds, isolating interference, estimating Direction of Arrival (DoA), and clustering interference samples into epochs to precisely remove interference while preserving target information.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Object-affected harmful factors

If time-frequency thresholding and zeroing processes are employed to remove interference signals, then interference removal is improved, but target signals are also removed resulting in reduced signal-to-noise ratio and ambiguous sidelobes

Engineering Contradiction:
Improveinterference removalVSAvoidsignal-to-noise ratio
Core Design Contradiction:
Object-affected harmful factorsVSReliability

Solution Approach 1:

The patent segments the time-frequency spectrum into multiple frequency bins and applies frequency-specific suppression thresholds to each bin. This allows selective removal of interference in certain frequency regions while preserving target signals in other frequency regions, avoiding the blanket zeroing approach that degrades overall signal quality.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent applies different suppression thresholds locally to different frequency bins based on the statistical characteristics of samples in each bin. This local adaptation allows the system to aggressively suppress interference in frequency regions where it dominates while using more conservative thresholds in regions where target signals are present, thereby maintaining signal-to-noise ratio.

Inventive Principle:
Principle #3Local quality

2Object-affected harmful factors

If zeroing patterns are applied to remove interference, then interference suppression is improved, but quasi-random zeroing results in raised sidelobe floor in range and Doppler spectrum

Engineering Contradiction:
Improveinterference suppressionVSAvoidsidelobe floor
Core Design Contradiction:
Object-affected harmful factorsVSMeasurement precision

Solution Approach 1:

The patent segments the spectrum into frequency bins and applies structured suppression patterns within each bin rather than random zeroing. This segmentation allows control over the spatial distribution of zeroed samples, preventing the random phase relationships that cause elevated sidelobe floors.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent changes the suppression threshold parameter dynamically for each frequency bin based on statistical analysis of the samples. This adaptive parameter adjustment creates a more regular and predictable suppression pattern compared to fixed or random zeroing, thereby maintaining better sidelobe characteristics.

Inventive Principle:
Principle #35Parameter changes

3Device complexity

If removed interference spectrogram samples are discarded, then processing complexity is reduced, but useful information about interference targets is lost

Engineering Contradiction:
Improveprocessing complexityVSAvoidinterference target information
Core Design Contradiction:
Device complexityVSLoss of information

Solution Approach 1:

The patent converts the harmful interference samples into beneficial information by analyzing their statistical characteristics to determine suppression thresholds. The same samples that contain interference also contain information about the interference sources, and by statistically characterizing them, the system extracts useful data about interference targets while still achieving suppression.

Inventive Principle:
Principle #22Blessing in disguise (Convert harm into benefit)

Solution Approach 2:

The patent introduces statistical analysis as an intermediary process between receiving the interference samples and discarding them. This intermediary step extracts meaningful information from the interference samples about target locations and characteristics before the samples are used for threshold determination, thereby preserving useful information that would otherwise be lost.

Inventive Principle:
Principle #24Intermediary (Mediator)

Data Source

PatentEP4184197B1Automotive radar with time-frequency-antenna domain threshold interference isolation and localization fusion
Publication Date: 2026.03.04 NXP BV
  • EP4184197B1 patent drawingFigure 1A
  • EP4184197B1 patent drawingFigure 1B
  • EP4184197B1 patent drawingFigure 2

AI summary

Described are method and systems that implement time frequency domain threshold interference and localization fusion to resolve interference issues in an automotive radar system, that produces spectrograms using Short-Time Fourier Transform (STFT) for all receiving antennas of the automotive radar system. For each STFT frequency a suppression threshold is determined. Interference is isolated for each STFT frequency by removing the interference from samples that are above the suppression threshold by using a filter. Direction of Arrival (DoA) is estimated for each interference spectrogram cell using measurements from all the receiving antennas. Interference samples are clustered using the DoA into epochs of chirps.