Radar Data Compression Using Cross-Correlation Reconstruction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Modern radar systems face challenges in efficiently storing and processing large volumes of radar data due to high memory requirements, particularly in automotive applications, while preserving critical information for accurate direction-of-arrival estimation and velocity measurement.
Innovation Solution
A data compression technique that selectively stores a subset of radar data samples and uses cross-correlation to reconstruct omitted samples, maintaining phase information and reducing memory footprint by leveraging a Toeplitz matrix for efficient computation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the entire radar data cube is stored to preserve all relevant information, then measurement precision and reliability are maintained, but memory consumption increases significantly
Solution Approach 1:
The patent extracts and stores only a subset of radar data samples (e.g., every other sample or samples at specific time instances) rather than the entire data cube. This selective extraction reduces memory consumption while the missing samples are reconstructed using interpolation techniques that leverage the correlation between adjacent samples, thereby maintaining measurement precision for direction-of-arrival estimation and velocity measurement.
Solution Approach 2:
The patent changes the sampling parameter by storing samples at reduced density (e.g., reducing the sampling rate by a factor of 2 or more). Instead of storing all N samples, only N/K samples are stored where K>1. The missing samples are recovered through interpolation methods that exploit the temporal and spatial correlation in radar data, allowing the system to maintain accuracy while reducing memory requirements by a factor of K.
2Quantity of substance
If data compression techniques are applied to reduce memory usage, then memory consumption decreases, but phase information may be lost or reconstruction becomes complex
Solution Approach 1:
The patent employs self-service interpolation where the missing radar samples are reconstructed using only the stored samples and their inherent correlations. The interpolation process automatically recovers phase information by leveraging the continuous phase evolution characteristics of radar signals. This self-service approach eliminates the need for external phase information or complex reconstruction algorithms, thereby reducing memory consumption without losing critical phase data needed for accurate DoA and velocity estimation.
3Measurement precision
If more antennas are integrated to increase data resolution, then measurement precision improves, but memory requirements and device complexity increase proportionally
Solution Approach 1:
The patent applies segmentation to the radar data cube by dividing it into manageable segments along the fast-time dimension. Instead of storing and processing the entire high-resolution data cube from multiple antennas simultaneously, the data is segmented into smaller chunks that can be stored and processed separately. This segmentation reduces the peak memory requirements and device complexity while maintaining the overall high resolution through coherent integration of the segmented data during signal processing.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Methods and systems are provided for efficient radar data compression and reconstruction. A radar processing unit receives and digitizes radar signals reflected from one or more objects, storing a first set of indexed samples while omitting a second set from storage. Sets of cross-correlation values are calculated between the second set of samples and the first set of samples within a local sliding window context. These values are used to reconstruct the second set of samples. The reconstructed and stored samples are then utilized to generate a radar map (such as a range-velocity map, range-Doppler map, or range-angle map, depending on the radar-cube axis of compression).