Radar Super-Resolution via Neural Network Data Cube Processing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Radar systems face limitations in resolution due to size, cost, and weight constraints, making it challenging to achieve high detection accuracy for objects in various applications.
Innovation Solution
The method employs deep learning techniques by obtaining first-resolution time samples from frequency-modulated signals, reducing resolution, implementing a matched filter, and processing data with a neural network to achieve super-resolution, allowing for higher accuracy in object detection using a second-resolution radar system.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the radar system uses more transmit elements and receive elements to improve resolution, then measurement precision is improved, but device complexity and cost increase
Solution Approach 1:
The patent creates a virtual high-resolution radar system by copying and processing low-resolution data through neural network algorithms. Instead of physically adding more transmit and receive elements, the system generates virtual signals and data cubes that simulate what a high-resolution system would produce, thereby achieving high measurement precision without increasing device complexity
Solution Approach 2:
The patent replaces the mechanical approach of adding more physical radar elements with a computational approach using neural networks and signal processing algorithms. The system substitutes physical hardware expansion with software-based super-resolution techniques, including data cube generation, neural network processing, and virtual signal synthesis to achieve enhanced detection resolution
2Measurement precision
If the radar system increases the number of transmit and receive elements to improve resolution, then measurement precision is improved, but size and weight increase
Solution Approach 1:
The patent generates virtual data cubes and signals that copy the characteristics of high-resolution radar data without requiring the physical infrastructure. By creating virtual representations of high-resolution measurements through algorithmic processing of low-resolution inputs, the system achieves high detection precision while maintaining a compact radar system size
Solution Approach 2:
The patent transitions from physical dimension expansion (adding more elements) to computational dimension expansion (creating multi-dimensional data cubes with range, Doppler, and angular dimensions). The system uses four-dimensional data cubes and neural network processing in parameter space to achieve high resolution without increasing physical volume
3Measurement precision
If the radar system uses more transmit and receive elements to improve resolution, then measurement precision is improved, but cost increases
Solution Approach 1:
The patent creates virtual high-resolution radar data by copying and processing low-resolution measurements through neural network algorithms. This approach eliminates the need to manufacture and deploy expensive high-resolution hardware, achieving high measurement precision through software-based super-resolution instead of costly hardware upgrades
Solution Approach 2:
The patent changes the processing parameters and data representation (creating four-dimensional data cubes with enhanced range, Doppler, and angular dimensions) rather than changing physical hardware parameters. By transforming low-resolution inputs into high-resolution outputs through parameter-space manipulation and neural network processing, the system achieves high detection precision without the cost of high-resolution hardware
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach enhances the detection resolution of radar systems, enabling more precise object detection and tracking in applications like vehicle automation without the need for costly or bulky hardware upgrades.
Implementation Method 1
transmissions by a first-resolution radar system of multiple frequency-modulated signals
Implementation Method 2
obtain first-resolution time samples from reflections based on transmissions
Implementation Method 3
implementing a matched filter on the first-resolution time samples to obtain a first-resolution data cube
Implementation Method 4
processing the second-resolution data cube with a neural network to obtain a third-resolution data cube
Implementation Method 5
deep learning for super resolution in a radar system
Data Source
AI summary
A system and method to use deep learning for super resolution in a radar system include obtaining first-resolution time samples from reflections based on transmissions by a first-resolution radar system of multiple frequency-modulated signals. The first-resolution radar system includes multiple transmit elements and multiple receive elements. The method also includes reducing resolution of the first-resolution time samples to obtain second-resolution time samples, implementing a matched filter on the first-resolution time samples to obtain a first-resolution data cube and on the second-resolution time samples to obtain a second-resolution data cube, processing the second-resolution data cube with a neural network to obtain a third-resolution data cube, and training the neural network based on a first loss obtained by comparing the first-resolution data cube with the third-resolution data cube. The neural network is used with a second-resolution radar system to detect one or more objects.