MIMO Radar Matrix-Rank Estimation for Multiple Objects
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional automotive radar systems using multiple-input multiple-output (MIMO) technology face challenges in accurately estimating the number of detected objects due to multiple directions of arrival at a given range-Doppler detection cell, complicating downstream processing and decreasing detection accuracy, and often require complex angle finding algorithms tailored to specific antenna array configurations.
Innovation Solution
A MIMO radar system processes received echoes to generate a matrix, determining its rank to estimate the number of detected objects without relying on angle finding algorithms or antenna array calibration, utilizing a neural network accelerator to compute the matrix rank.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional angle finding algorithms are used to estimate the number of objects, then detection accuracy can be maintained, but the device complexity and processing requirements increase significantly
Solution Approach 1:
The patent extracts the essential information needed for object quantity estimation by forming a matrix from receive vectors and computing its rank, eliminating the need for complex angle finding algorithms. This extraction approach focuses only on the necessary data (matrix rank) rather than performing full angular analysis, thereby reducing computational complexity while maintaining estimation accuracy.
Solution Approach 2:
The patent replaces the mechanical/algorithmic angle finding process with a mathematical matrix rank computation. Instead of using iterative angle estimation algorithms tailored to specific antenna configurations, the system substitutes this with a universal matrix rank calculation that directly yields object quantity information, significantly simplifying the processing system.
2Measurement precision
If angle finding algorithms tailored to specific antenna array configurations are used, then detection accuracy is maintained, but the adaptability to different configurations decreases
Solution Approach 1:
The patent creates a universal object quantity estimation method that works with any MIMO radar antenna array configuration. By formulating the estimation problem in terms of matrix rank computation from receive vectors, the system achieves configuration independence - the same mathematical approach applies regardless of the specific number or arrangement of transmit and receive antennas, thereby providing multi-functional adaptability.
Solution Approach 2:
The patent changes the fundamental parameter used for object quantity estimation from angular information (which is configuration-dependent) to matrix rank (which is configuration-independent). This parameter transformation allows the system to maintain detection accuracy while becoming adaptable to various antenna array configurations without requiring re-tailoring of algorithms.
3Device complexity
If multiple directions of arrival are considered at each range-Doppler cell, then detection accuracy decreases due to processing complexity, but ignoring them loses valuable information
Solution Approach 1:
The patent extracts the essential object quantity information from the multiple directions of arrival data through matrix formation and rank computation. Instead of processing each direction separately (which increases complexity), the system extracts the collective information content by constructing a matrix from all receive vectors and computing its rank, thereby capturing all directional information efficiently without excessive processing complexity.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Devices and methods for object quantity estimation in a radar system include receiving a plurality of radar echoes at a plurality of reception antennas of the radar system. The plurality of radar echoes correspond to radar signals transmitted from multiple transmission antennas of the radar system. A radar processor of the radar system generates a matrix based on the plurality of radar echoes and estimates the number of objects based on computing a rank of the matrix.