FMCW Radar FFT Processing With Non-Coherent Accumulation

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Conventional FMCW radars require large amounts of memory for FFT processing, leading to increased size and power consumption, and existing memory compression techniques result in accuracy degradation.

Innovation Solution

Implement a radar apparatus with multiple receivers and a processor that performs non-coherent accumulation of FFT data across multiple chirps, using nonlinear operations to generate a two-dimensional image, reducing memory requirements by storing output data in a secondary memory.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If conventional FFT processing is used in FMCW radar, then accurate position and velocity of obstacles can be obtained, but large amounts of memory are required in the processor

Engineering Contradiction:
Improveaccuracy of obstacle detectionVSAvoidmemory requirement
Core Design Contradiction:
Measurement precisionVSQuantity of substance

Solution Approach 1:

The patent divides the FFT processing into multiple frames, where each frame processes a subset of chirps. The processor performs FFT on individual frames separately and then combines results through non-coherent accumulation, rather than processing all chirps simultaneously in one large FFT. This segmentation reduces the memory required for each individual FFT operation while maintaining overall detection accuracy through the combination of multiple frame results.

Inventive Principle:
Principle #1Segmentation

2Quantity of substance

If memory compression techniques are used to reduce memory size, then memory requirements decrease, but accuracy of the radar system degrades

Engineering Contradiction:
Improvememory sizeVSAvoidradar accuracy
Core Design Contradiction:
Quantity of substanceVSMeasurement precision

Solution Approach 1:

The patent extracts only the essential information from each frame's FFT processing - specifically the magnitude spectrum or peak information - and discards the full complex data. By taking out only the necessary components for obstacle detection and combining these extracted features across frames through non-coherent accumulation, the system achieves accurate detection with significantly reduced memory requirements, avoiding the need to store complete complex FFT data from all chirps.

Inventive Principle:
Principle #2Taking out (Extraction)

3Quantity of substance

If multiple frames are transmitted for single viewing to reduce memory, then power consumption increases and real-time application becomes difficult

Engineering Contradiction:
Improvememory requirementVSAvoidpower consumption
Core Design Contradiction:
Quantity of substanceVSUse of energy by moving object

Solution Approach 1:

The patent implements a dynamic processing approach where the number of frames and chirps per frame can be adjusted based on operational requirements. The system dynamically balances between processing multiple frames for memory reduction and maintaining real-time performance by optimizing the frame structure and accumulation process, allowing flexible adaptation to different power and performance constraints.

Inventive Principle:
Principle #15Dynamics

4Adaptability or versatility

If large memory is allocated for FFT processing data, then complete range viewing in single frame is enabled, but device size increases

Engineering Contradiction:
Improverange viewing capabilityVSAvoidradar size
Core Design Contradiction:
Adaptability or versatilityVSVolume of stationary object

Solution Approach 1:

The patent transitions from a single-dimension approach (storing all range data in one large memory array) to a multi-dimensional approach by organizing data across multiple frames and using non-coherent accumulation in the signal processing domain. This dimensional reorganization allows the system to achieve complete range viewing capability through temporal accumulation across frames rather than spatial storage in a single large memory, thereby reducing the physical size of the radar device.

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

Data Source

PatentEP3350617B1Method and apparatus for FMCW radar processing
Publication Date: 2026.01.28 TEXAS INSTRUMENTS INC
  • EP3350617B1 patent drawingFigure 1~3B
  • EP3350617B1 patent drawingFigure 2A~2B
  • EP3350617B1 patent drawingFigure 2C

AI summary

In described examples, a radar apparatus (100) includes a transmitter (101) that transmits a first chirp. The first chirp is scattered by one or more obstacles to generate scattered signals. A plurality of receivers (110) receives the scattered signals. Each of the receivers (110) generates a digital signal in response to one of the scattered signals. A processor (120) is coupled to the receivers (110) and receives the digital signals from the receivers (110). The processor (120) performs range FFT (fast Fourier transform) and angle FFT on the digital signals received from the receivers (110) to generate a first matrix of complex samples.