RF Signal Image Generation With Automatic Bounding Boxes

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Manual labeling of RF signal images is cumbersome and time-consuming, especially when dealing with large volumes or complex RF signals with superimposed impairments.

Innovation Solution

A system that automatically generates images of RF signals and/or impairments, including bounding boxes to indicate signal and impairment locations, eliminating the need for manual labeling.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If manual labeling is used to label RF signal images, then labeling accuracy can be ensured through expert knowledge, but the process becomes cumbersome and time-consuming

Engineering Contradiction:
Improvelabeling accuracyVSAvoidlabeling time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The system performs preliminary action by automatically generating bounding boxes and labels during the image generation process itself, before any manual intervention is needed. The RF signal generator creates signals with known parameters, and the calculation unit immediately computes bounding boxes based on these parameters, eliminating the need for subsequent manual labeling while maintaining accuracy through the use of precise signal generation models

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The system applies self-service by enabling the RF signal generation system to automatically label its own output images without external human intervention. The calculation unit uses the known signal parameters from the RF signal generator to automatically calculate and attach bounding boxes to the generated spectrogram images, making the system self-labelling and eliminating time-consuming manual processes

Inventive Principle:
Principle #25Self-service

2Measurement precision

If manual labeling is used for complex RF signals with superimposed impairments, then accurate identification of specific signals can be achieved, but the complexity and time required increase significantly

Engineering Contradiction:
Improvesignal identification accuracyVSAvoidlabeling process complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The system performs preliminary action by pre-defining signal parameters and characteristics during the RF signal generation phase. Before the complex labeling task arises, the system has already generated signals with known parameters including impairments, and the calculation unit is prepared to automatically compute bounding boxes based on these pre-established parameters, simplifying what would otherwise be a complex manual identification process

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The system applies self-service by automatically handling the complex task of identifying and labeling superimposed signals with impairments. The calculation unit uses the known signal parameters from the RF signal generator to automatically distinguish and bound multiple superimposed signals in the spectrogram, eliminating the need for expert manual analysis while managing the complexity through algorithmic processing

Inventive Principle:
Principle #25Self-service

3Quantity of substance

If a large amount of labeled RF images is required, then comprehensive training data can be obtained, but the manual labeling effort becomes prohibitively time-consuming

Engineering Contradiction:
Improvevolume of training dataVSAvoidtotal labeling time
Core Design Contradiction:
Quantity of substanceVSLoss of time

Solution Approach 1:

The system performs preliminary action by integrating the labeling function into the image generation process itself. As each RF signal image is generated, the calculation unit immediately computes and attaches the corresponding bounding boxes and labels, so that every generated image comes pre-labeled. This preliminary automatic labeling eliminates the need for subsequent manual processing of large volumes of images

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The system applies self-service by automatically generating large volumes of labeled training data without requiring external human effort. The RF signal generator can produce numerous signals with varying parameters, and the calculation unit automatically labels each one, enabling the system to self-generate comprehensive training datasets at scale while eliminating the prohibitively time-consuming manual labeling that would otherwise be required

Inventive Principle:
Principle #25Self-service

Data Source

PatentEP4597885A1System and method for generating images of RF signals and/or RF impairments
Publication Date: 2025.08.06 ROHDE & SCHWARZ GMBH & CO KG
  • EP4597885A1 patent drawingFigure 1
  • EP4597885A1 patent drawingFigure 2
  • EP4597885A1 patent drawingFigure 3

AI summary

The disclosure relates to a system (10) for generating images of RF signals and/or RF impairments. The system (10) comprises: a receiving unit (11) configured to receive information on a number of RF signals and/or RF impairments, and image parameters; one or more RF signal generators (12) configured to generate the number of RF signals and/or RF impairments according to the received information; an image generation unit (13) configured to generate an image of a frequency spectrum and/or a spectrogram of the generated RF signals and/or RF impairments according to the received image parameters; a calculation unit (14) configured to calculate a respective bounding box for at least one of the RF signals and/or for at least one of the RF impairments in the generated image, wherein the bounding box indicates a location of the at least one RF signal and/or the at least one RF impairment in the generated image; and a memory (15) configured to store the generated image together with the bounding box(es).