Automotive Radar AoA Estimation with SBX Support Set Refinement
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional automotive radar systems face challenges in accurately estimating the angle of arrival (AoA) of objects due to sparsity in the angular domain, leading to spurious sidelobes and grating lobes, and existing algorithms like MP and OMP are sensitive to antenna array geometry and hyperparameter settings, resulting in suboptimal results.
Innovation Solution
The implementation of a Single Best Exchange (SBX) algorithm for AoA estimation in automotive radar systems, which performs exchange operations on selected and unselected supports to optimize the support set, reducing spurious identifications and improving accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional algorithms (MP, OMP) are used for AoA estimation, then the system can perform basic angle estimation, but the results are sensitive to antenna array geometry and hyperparameter settings, leading to suboptimal accuracy
Solution Approach 1:
The patent transforms the AoA estimation problem into a sparse linear regression problem with L0 norm regularization, changing the mathematical formulation to be more robust to geometric variations. The algorithm dynamically adjusts the regularization parameter lambda and iteratively refines the support set, making the solution less sensitive to initial geometric configurations and hyperparameter choices.
Solution Approach 2:
The patent implements an iterative algorithm that uses feedback from the measurement vector and steering vector matrix to progressively refine the support set. The algorithm continuously evaluates the residual error and adjusts the selected supports based on this feedback, converging to a solution that is robust to geometric variations and hyperparameter settings.
2Measurement precision
If existing algorithms are used for AoA estimation, then processing can be performed, but spurious sidelobes and grating lobes are generated, reducing detection accuracy
Solution Approach 1:
The patent extracts and eliminates spurious components by identifying and removing incorrect supports from the solution set. Through the iterative refinement process, the algorithm distinguishes between true signal components and spurious sidelobes, extracting only the valid AoA estimates while discarding harmful artifacts.
Solution Approach 2:
The patent converts the harmful spurious sidelobes into useful information by using the residual error signal to identify and correct incorrect supports. The presence of spurious components actually guides the iterative algorithm to refine its solution, transforming the harmful artifacts into indicators for improvement.
3Measurement precision
If iterative algorithms with multiple operations are used for AoA estimation, then accuracy can be improved, but computational overhead increases
Solution Approach 1:
The patent applies partial action by performing exchange operations only when necessary to improve the solution. Instead of exhaustively searching all possible support combinations, the algorithm makes selective exchanges based on the residual error and steering vector correlations, achieving good accuracy with reduced computational effort.
Solution Approach 2:
The patent introduces dynamic adaptability in the exchange operation process. The algorithm dynamically adjusts the exchange strategy based on the current residual error and the correlation between steering vectors, making the computational process more efficient by focusing on the most promising exchanges rather than performing uniform operations.
Data Source
AI summary
A radar system includes, transmitters, receivers, and a controller that determines a measurement vector using signals received by the plurality of receiver modules, determines a steering vector matrix, and determines a plurality of supports using the measurement vector. The controller executes a regression algorithm to determine a weight vector that defines a relationship between the measurement vector and the steering vector matrix by defining a set of selected supports out of the plurality of supports, executes an exchange operation to determine an optimized set of selected supports by removing a first support from the set of selected supports and adding a second support to the set of selected supports, and calculates the weight vector using the optimized set of selected supports. The controller is configured to determine an estimated angle of arrival of a first object by correlating the steering vector matrix to the measurement vector using the weight vector.