Composite Signal Decomposition Using Gaussian Process Envelopes
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing signal processing methods face challenges in accurately decomposing composite signals, such as those produced by sensors monitoring vital signs, into individual signals like heart beats and respiration, due to the composite nature of the signals.
Innovation Solution
The method involves determining upper and lower envelopes for a composite signal based on a smoothness parameter using Gaussian processes, allowing for the derivation of a smooth estimate that is more accurate than legacy approaches, and subsequently isolating transient components by comparing the smooth estimate with the composite signal.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional signal processing methods are used to decompose composite signals, then the processing is simpler, but the decomposition accuracy is lower
Solution Approach 1:
The patent segments the composite signal decomposition process into two distinct stages: first extracting the smooth component using envelope detection and Gaussian processes, then isolating the transient component by subtraction. This segmentation allows each component to be processed with appropriate methods, improving overall decomposition accuracy while maintaining manageable complexity.
Solution Approach 2:
The patent introduces Gaussian processes as an intermediary mathematical tool to model the smooth component of the composite signal. This intermediary approach enables accurate separation of smooth and transient components by providing a probabilistic framework that captures the underlying signal structure without requiring direct complex decomposition of the entire composite signal.
2Measurement precision
If envelope detection methods are used to extract smooth components, then the smooth component can be identified, but the precision is insufficient compared to Gaussian process approaches
Solution Approach 1:
The patent changes the mathematical parameters and assumptions used in smooth component estimation by transitioning from deterministic envelope detection to probabilistic Gaussian process modeling. This parameter change allows incorporation of prior knowledge about signal smoothness and uncertainty quantification, significantly improving estimation accuracy despite increased computational complexity.
Data Source
AI summary
Disclosed herein are example systems and approaches for decomposition of composite signals. Decomposition of the composite signals may include derivation of two or more signals from the composite signals. An upper envelope and lower envelope may be determined for a composite signal in accordance with a smoothness parameter. A smooth estimate may be produced based on the upper envelope and the lower envelope, where the smooth estimate provides an estimate for a smooth component of the composite signal, which may be more accurate than legacy approaches.