Automotive Radar CFAR Pre-Processing for Low-SNR Target Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing CFAR receiver processing solutions struggle to efficiently and accurately detect multiple radar targets with varying signal-to-noise ratios (SNR) due to computational complexity and the masking effect of high SNR targets, especially in automotive radar systems with limited resources.
Innovation Solution
A radar system and method that employs non-coherent integration to estimate a precise CFAR threshold from low SNR data, which is then applied to coherently integrated high SNR data, using a scaling factor to maintain a constant false alarm rate and enhance detection probability without degrading false alarm rates.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional CFAR algorithms are used to detect radar targets, then the false alarm rate is controlled, but high SNR targets mask low SNR targets reducing detection probability
Solution Approach 1:
The patent segments the detection process into two distinct integration paths: non-coherent integration for threshold estimation and coherent integration for target detection. This segmentation allows the system to estimate the noise floor using non-coherent integration (which is robust to high SNR targets) while maintaining high detection sensitivity through coherent integration, thereby preventing high SNR targets from masking low SNR targets while still controlling false alarm rates.
Solution Approach 2:
The patent introduces non-coherent integration as an intermediary process that creates a separate noise floor estimation path. This intermediary mechanism allows the system to determine the CFAR threshold based on non-coherent integration results, which are not affected by high SNR target masking, and then apply this threshold to coherent integration results for final target detection, thus resolving the masking problem while maintaining false alarm control.
2Measurement precision
If computationally exhaustive super-resolution radar detection algorithms are applied to entire raw data cube, then high resolution Direction of Arrival estimations are obtained, but real-time processing capabilities are limited
Solution Approach 1:
The patent applies preliminary CFAR detection to the raw data cube before applying super-resolution algorithms. By first using computationally efficient CFAR algorithms to identify candidate target locations and then applying high-resolution Direction of Arrival estimation only to these detected targets rather than the entire data cube, the system achieves high measurement precision while maintaining real-time processing capability.
3Measurement precision
If CFAR threshold is increased to detect low SNR targets, then detection probability improves, but high SNR targets cause threshold elevation masking adjacent targets
Solution Approach 1:
The patent segments the threshold estimation process from the detection process by using non-coherent integration for threshold estimation and coherent integration for detection. This segmentation ensures that the CFAR threshold is determined based on non-coherent integration results that are not influenced by high SNR targets, preventing threshold elevation and the subsequent masking of adjacent targets while still enabling detection of low SNR targets.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A vehicle radar system, apparatus and method use a radar control processing unit generate compressed radar data signals, to apply the compressed radar data signals in parallel as a three-dimensional matrix to a coherent integrator (which generates a two-dimensional matrix of coherently integrated image data) and a non-coherent integrator (which generates a two-dimensional matrix of non-coherently integrated image data), and to generate a constant false alarm rate (CFAR) threshold from the two-dimensional matrix of non-coherently integrated image data for application to the two-dimensional matrix of coherently integrated image data to detect one or more targets in the MIMO radar signal returns from sample values from the two-dimensional matrix of coherently integrated image data that exceed the CFAR threshold.