Linear Response Extraction for ML Waveform Measurement
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current machine learning techniques for test and measurement systems are inefficient due to reliance on short pattern waveforms, which only appear in a portion of the whole data pattern, leading to complications in obtaining sufficient data and losing time sequence information, especially when using equalizers for complex measurements.
Innovation Solution
The use of an extracted linear fit pulse approach that utilizes all samples in the waveform, providing time sequence information and improving data efficiency by representing data as 1-D or 2-D vectors for neural networks, allowing for more accurate equalization parameter determination and faster measurement processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If short pattern waveforms are used for machine learning measurements, then the measurement process can be simplified, but the time sequence information is lost and sufficient data cannot be obtained
Solution Approach 1:
The patent extracts the linear response component from the received signal by processing the entire waveform data. This extraction process separates the useful linear response information from the complete signal, allowing machine learning models to utilize time sequence information without being overwhelmed by the full complexity of the original waveform, thus resolving the contradiction between simplification and information preservation
Solution Approach 2:
The patent transforms the temporal waveform data into a different representation by extracting linear response characteristics. This dimensional transformation converts the time-domain signal into a form that preserves essential time sequence information while making it suitable for machine learning processing, effectively resolving the information loss problem while maintaining computational simplicity
2Device complexity
If short pattern waveforms are used for machine learning, then data processing can be reduced, but data efficiency decreases and training time increases
Solution Approach 1:
By extracting the linear response from the complete waveform, the patent obtains a concentrated representation that contains all necessary time sequence information. This extracted representation is more efficient for training machine learning models compared to using fragmented short patterns, thereby improving data efficiency and reducing training time while maintaining processing simplicity
Solution Approach 2:
The patent processes the entire continuous waveform to extract the linear response, ensuring that no useful time sequence information is lost. This continuous processing approach maximizes data efficiency by utilizing all available information from the signal, leading to faster and more accurate model training compared to discontinuous short pattern sampling
Data Source
AI summary
A test and measurement instrument has one or more ports configured to receive a signal one or more devices under test (DUT), and one or more processors configured to execute code that causes the one or more processors to: acquire a waveform from the signal, derive a pattern waveform from the waveform, perform linear response extraction on the pattern waveform, present one or more data representations including a data representation of the extracted linear response to a machine learning system, and receive a prediction for a measurement from the machine learning system. A method of performing a measurement on a waveform includes acquiring the waveform at a test and measurement device, deriving a pattern waveform from the waveform, performing linear response extraction on the pattern waveform, presenting one or more data representations including a data representation of the extracted linear response to a machine learning system, and receiving a prediction of the measurement from the machine learning system.