Adaptive Multi-Trace Carving for Frequency Estimation in Low SNR Signals
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing technologies face challenges in accurately detecting and estimating frequency components from signals, particularly in low signal-to-noise ratio (SNR) conditions and when dealing with multiple quasiperiodic sources.
Innovation Solution
A method and apparatus for tracking candidate frequency traces from a time-frequency representation of a signal, using techniques such as convolutional neural networks and adaptive multi-trace carving, to identify and output an estimated frequency vector under noisy conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If traditional estimation algorithms are applied individually to each temporal segment, then the algorithm complexity is reduced and ease of operation is improved, but measurement precision deteriorates due to inability to exploit temporal correlation
Solution Approach 1:
The signal is divided into multiple temporal segments, and frequency estimation is performed on each segment separately using traditional algorithms. This segmentation allows the system to maintain algorithmic simplicity while processing manageable portions of the signal independently.
Solution Approach 2:
The frequency estimates from multiple temporal segments are combined through temporal correlation analysis. By merging the results and exploiting the correlation between neighboring segments, the system achieves improved frequency estimation accuracy that overcomes the limitations of individual segment analysis.
2Measurement precision
If subspace methods such as MUSIC and ESPRIT are used, then measurement precision is improved through parametric models, but device complexity increases due to building pseudo power spectra
Solution Approach 1:
The patent extracts the essential frequency information from each temporal segment using traditional estimation algorithms, separating the frequency extraction task from the complex pseudo power spectrum building process required by subspace methods. This extraction approach maintains accuracy while reducing computational complexity.
Solution Approach 2:
The patent replaces the mechanical process of building pseudo power spectra (required by MUSIC and ESPRIT) with a simpler statistical approach that uses temporal correlation of frequency estimates. This substitution eliminates the need for complex parametric modeling while achieving comparable or superior accuracy in certain conditions.
3Ease of operation
If frame-wise estimation algorithms are used, then ease of operation is improved, but reliability deteriorates as SNR drops and outliers are generated
Solution Approach 1:
The system performs preliminary frequency estimation on each temporal segment separately, obtaining candidate frequency values before combining them. This preliminary action allows for subsequent filtering and validation steps that remove outliers and improve reliability in low SNR conditions.
Solution Approach 2:
The patent implements feedback mechanisms where frequency estimates from neighboring segments are used to validate and correct estimates from the current segment. By comparing temporal correlations and identifying consistent patterns across segments, the system filters out outliers and maintains high reliability even when individual segment estimates are noisy.
Data Source
AI summary
Systems, methods, apparatuses, and computer program products for tracking weak signal traces under severe noise and/or distortions. A method may include tracking at least one candidate frequency trace from a time-frequency representation of a signal. The method may also include identifying a frequency trace of the signal based on tracking results. In addition, the method may include outputting an estimated frequency vector related to the frequency trace. Further, the tracking may be performed under a noisy condition environment.