Blind Source Separation Using 2-D Histograms for Low-SNR Signals
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing blind source separation systems struggle to detect and separate low-SNR signals in real-time due to their reliance on high signal-to-noise ratios and inability to distinguish signals from noise, particularly when signal power is near or below the background noise level.
Innovation Solution
The proposed system uses a two-dimensional stochastic histogram to gather wideband signal power spectral information without specialized Fourier transform hardware or large memory, matching filter characteristics to signal characteristics, and employing persistent barcodes to drive unused filters and enhance signal detection and separation by combining real-time and long-term information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional blind source separation filters are used, then high-SNR signals can be separated effectively, but low-SNR signals near or below background noise level cannot be detected
Solution Approach 1:
The patent transforms the one-dimensional frequency analysis into a two-dimensional frequency-time histogram analysis. By accumulating signal energy across multiple time samples in a frequency histogram, low-SNR signals that are indistinguishable from noise in single samples can be detected through their cumulative energy distribution across the frequency-time plane.
Solution Approach 2:
The system performs preliminary accumulation of signal energy in the frequency histogram before making detection decisions. This preliminary action of gathering energy information across multiple time samples allows the system to build up detectable signal energy from low-SNR inputs that would otherwise be lost in noise.
2Measurement precision
If signal energy accumulation over long periods is used to detect low-SNR signals, then detection sensitivity improves, but real-time processing capability deteriorates
Solution Approach 1:
The frequency histogram is updated dynamically with each new time sample, allowing the system to adapt to changing signal conditions in real-time. The histogram structure enables continuous accumulation of energy information without requiring storage of entire signal records, maintaining processing efficiency while improving detection sensitivity.
Solution Approach 2:
The system accumulates energy information partially - only the essential frequency and time components are stored in the histogram rather than complete signal waveforms. This partial accumulation provides sufficient information for low-SNR detection while avoiding the computational burden of processing complete long-term signal records.
3Adaptability or versatility
If more BSS filter channels are used to cover more frequency ranges, then signal separation coverage improves, but system complexity and computational load increase
Solution Approach 1:
The frequency histogram serves multiple functions simultaneously: it acts as a signal detector, a frequency analyzer, and a time-correlation tool. This multi-functional approach allows the system to achieve comprehensive frequency coverage and low-SNR detection without proportionally increasing hardware complexity, as the same histogram structure performs multiple analytical tasks.
Data Source
AI summary
A blind source separation (BSS) system comprises means for gathering wideband signal power spectral information without using special Fourier transform hardware/software systems and without using extremely large signal memories. The method involves taking the combined energy output of a large set of BSS filters that are operating under a blind source separation algorithm that looks for signals of interest, along with instantaneous filter characteristics such as center frequency, bandwidth and spectral response, and records a weighted spread energy into a frequency- and time-based histogram. This information can then be used to compare against the BSS signal output and signal energy which does not correspond to existing signal output is added to additional BSS signal output. This system can operate in real time, but uses long-term averaging to enhance signal detection.