AI Signal Analysis Using Bicoherence for Jitter Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current test and measurement systems face challenges in accurately identifying and classifying jitter and other distortions in serial data communications, particularly in modern serial data communications where the serial data clock is not transmitted with the data, leading to data errors at the receiving end.
Innovation Solution
The system employs bicoherence analysis combined with artificial intelligence, using short-term fast Fourier transform computations and neural networks to identify and separate jitter from other distortions such as inter-symbol interference, by normalizing the bispectrum and applying machine learning to classify signal characteristics.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional test and measurement systems are used to analyze jitter in serial data communications, then basic jitter measurement can be achieved, but accuracy in identifying and classifying jitter from other distortions deteriorates
Solution Approach 1:
The patent segments the complex signal analysis problem into distinct components by separating jitter analysis from other distortion analysis. It uses bicoherence computation to specifically isolate and identify jitter components from the composite signal, allowing for more accurate jitter measurement without interference from other distortions like inter-symbol interference.
Solution Approach 2:
The patent introduces bicoherence computation as an intermediary technique between the raw signal and the final jitter measurement. This intermediary method transforms the signal into a domain where jitter can be specifically identified and measured, acting as a mediator that enhances measurement precision while maintaining reliability.
2Measurement precision
If bicoherence computation is used to separate jitter from other distortions, then measurement precision improves, but device complexity increases
Solution Approach 1:
The patent replaces complex mechanical or manual signal analysis methods with automated computational algorithms. By using bicoherence computation and artificial intelligence, the system automatically separates and identifies jitter components without requiring manual intervention, thereby improving measurement precision while managing complexity through automation.
Solution Approach 2:
The patent changes the analysis domain parameters by transforming the time-domain signal into a bicoherence domain. This parameter transformation allows for more effective separation of jitter from other distortions, improving identification accuracy while the computational nature of the transformation manages the complexity through algorithmic processing.
3Measurement precision
If artificial intelligence and neural networks are applied to classify signal characteristics, then classification accuracy improves, but computational requirements and device complexity increase
Solution Approach 1:
The patent implements self-service through automated neural network classification. The system uses trained neural networks to automatically classify signal characteristics and identify jitter components without requiring manual analysis. This self-service approach improves classification accuracy while managing complexity through automation and pre-trained models.
Solution Approach 2:
The patent applies preliminary action by pre-training the neural networks with labeled data before deployment. This preliminary training phase allows the system to learn and store classification patterns in advance, so that during actual operation, the classification can be performed quickly and accurately without requiring complex real-time processing, thereby managing device complexity.
Data Source
AI summary
A test and measurement instrument for analyzing signals using machine learning. The test and measurement instrument can determine a recovered clock signal based on the digital signal, set window positions for a fast Fourier transform of the digital signal, window the digital signal into a series of windowed waveform data based on the window positions, transform each of the windowed waveform data into a frequency-domain windowed waveform data using a fast Fourier transform, and determine high-order spectrum data of each of the frequency-domain windowed waveform data. The test and measurement instrument includes a neural network configured to receive the high-order spectrum data of the frequency-domain windowed transform data and classify each windowed waveform data based on the high-order spectrum data.