Non-linear Signal Separation via Time Domain Projection

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Conventional non-linear state space projection methods for separating fetal electrocardiogram signals from a mother's body are inefficient due to high computational requirements and difficulty in accurately processing signals with low S/N ratios, especially when noise components change rapidly or are widespread, limiting their applicability in real-time measurements.

Innovation Solution

The method employs a time domain high-speed non-linear state space projection technique that stabilizes data by shifting cyclic signals and calculating summation averages, allowing for rapid noise removal and signal extraction without requiring extensive dimensional expansion or neighborhood definitions, thus reducing processing time significantly.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If conventional non-linear state space projection method is used to separate fetal electrocardiogram signals, then signal separation capability is improved, but processing time increases significantly (100 times slower)

Engineering Contradiction:
Improvesignal separation capabilityVSAvoidprocessing time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent segments the complex non-linear signal processing into distinct stages: cyclic signal identification, summation average calculation, and noise removal. This segmentation allows each stage to be optimized independently, with the summation average providing a stable reference that simplifies subsequent noise removal operations, thereby reducing overall processing time while maintaining separation capability

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent performs preliminary stabilization of the signal by calculating summation averages of cyclic components before conducting noise removal. This preliminary action creates a stable reference framework that simplifies the subsequent signal separation task, allowing faster processing in the final stage while maintaining high separation accuracy

Inventive Principle:
Principle #10Preliminary action

2Measurement precision

If high-dimensional state space projection is used to handle low S/N ratio signals, then signal extraction accuracy is improved, but computational complexity increases

Engineering Contradiction:
Improvesignal extraction accuracyVSAvoidcomputational complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent transforms the low S/N ratio signal into a higher-dimensional space by calculating summation averages across multiple cyclic periods. This dimensional transformation concentrates the signal energy while distributing noise, effectively improving S/N ratio without requiring complex high-dimensional state space projections, thus reducing computational complexity

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

Solution Approach 2:

The patent changes the temporal parameter by accumulating signals over multiple cyclic periods to form summation averages. This parameter transformation converts weak, noise-dominated individual cycles into a stable, high-S/N reference signal, improving extraction accuracy while avoiding the need for computationally intensive high-dimensional projections

Inventive Principle:
Principle #35Parameter changes

3Object-affected harmful factors

If conventional noise removal methods are used, then noise reduction is achieved, but signal quality deteriorates (all signals become dull)

Engineering Contradiction:
Improvenoise reductionVSAvoidsignal quality
Core Design Contradiction:
Object-affected harmful factorsVSMeasurement precision

Solution Approach 1:

The patent introduces summation averages of cyclic signals as an intermediary reference. This intermediary serves as a stable template that guides the noise removal process, allowing selective elimination of noise components while preserving the characteristic features of the fetal electrocardiogram signal, thus maintaining signal quality while achieving noise reduction

Inventive Principle:
Principle #24Intermediary (Mediator)

4Productivity

If real-time processing is implemented for clinical applications, then measurement speed is improved, but processing accuracy may deteriorate

Engineering Contradiction:
Improvemeasurement speedVSAvoidprocessing accuracy
Core Design Contradiction:
ProductivityVSMeasurement precision

Solution Approach 1:

The patent exploits the periodic nature of cardiac cycles by accumulating signals over multiple periods to form summation averages. This periodic approach allows real-time processing because each cycle contributes incrementally to the growing accuracy of the reference signal, enabling fast processing while maintaining high accuracy through the stabilizing effect of periodic accumulation

Inventive Principle:
Principle #19Periodic action

Data Source

PatentEP1935335B1Non-linear signal separation method using non-linear state space projection method
Publication Date: 2019.05.08 TOHOKU TECHNO ARCH CO LTD
  • EP1935335B1 patent drawingFigure 1~2
  • EP1935335B1 patent drawingFigure 3~4
  • EP1935335B1 patent drawingFigure 5~6

AI summary

There is provided a non-linear signal separation method using the non-linear state space projection method capable of separating an effective non-linear signal even if the S/N ratio is low by performing the time domain high speed non-linear state space projection when a signal is a multi-channel signal and has a periodicity. In the non-linear signal separation method using the non-linear state space projection method, an original signal having a complex signal which is a multi-channel and cyclic signal measured from one phenomenon is processed by using the time domain high-speed non-linear state space projection method so as to estimate a noise in the original signal and subtract the estimated noise from the original signal, thereby separating the signal to be measured in the original signal as a non-linear signal even when the S/N ratio is low.