Radar Data Compression Using Cross-Correlation Reconstruction

Resolve Bottlenecks,
Find Innovative Solutions
Generate 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

VSEngineering 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

Engineering Contradiction:
Improvedirection-of-arrival estimation accuracyVSAvoidmemory consumption
Core Design Contradiction:
Measurement precisionVSQuantity of substance

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.

Inventive Principle:
Principle #2Taking out (Extraction)

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.

Inventive Principle:
Principle #35Parameter changes

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

Engineering Contradiction:
Improvememory consumptionVSAvoidphase information
Core Design Contradiction:
Quantity of substanceVSLoss of information

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.

Inventive Principle:
Principle #25Self-service

3Measurement precision

If more antennas are integrated to increase data resolution, then measurement precision improves, but memory requirements and device complexity increase proportionally

Engineering Contradiction:
Improvedata resolutionVSAvoidmemory resources
Core Design Contradiction:
Measurement precisionVSDevice complexity

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.

Inventive Principle:
Principle #1Segmentation

Data Source

PatentEP4703767A1Radar data compression using hybrid spectral transformation
Publication Date: 2026.03.04 NXP BV
  • EP4703767A1 patent drawingFigure 1
  • EP4703767A1 patent drawingFigure 2
  • EP4703767A1 patent drawingFigure 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).