Short Pattern Waveform Database for ML Signal Measurement
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing machine learning systems for signal measurements in high-speed communication systems, such as those using eye diagrams, lack time sequence information, leading to inaccurate measurements due to differences between pre- and post-equalizer waveforms, especially in systems like PCIE Gen5 and IEEE 100G/400G Ethernet standards.
Innovation Solution
The use of short pattern waveform databases that contain time sequence information, allowing for faster and more accurate signal measurements by training machine learning systems on short patterns of different symbol lengths and applying these patterns to neural networks for analysis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If eye diagrams prior to equalization are used as inputs for machine learning measurements, then the measurement process can be simplified, but the measurement accuracy deteriorates because time sequence information is lost and post-equalizer waveforms differ significantly from pre-equalizer waveforms
Solution Approach 1:
The patent pre-generates a database of short pattern waveforms (e.g., PRBS7, PRBS15, PRBS31) that represent expected signal characteristics before equalization. During measurement, the system compares the actual captured waveform against this pre-prepared database to directly determine equalizer settings and measurements, eliminating the need for complex real-time analysis of pre-equalizer waveforms while maintaining accuracy aligned with post-equalizer performance
Solution Approach 2:
The patent creates simplified copies of complex post-equalizer waveform characteristics by storing representative short pattern waveforms in a database. Instead of analyzing complex time sequence data during measurement, the system uses these pre-captured waveform patterns as proxies, allowing accurate measurement determination without processing the full complexity of original signals
2Measurement precision
If full pattern waveforms are used for machine learning analysis, then measurement accuracy is maintained, but processing time increases significantly
Solution Approach 1:
The patent extracts only the essential measurement information from full pattern waveforms by capturing short pattern waveforms (specific time intervals containing critical signal characteristics). This extraction process isolates the most relevant portions of the waveform that contain sufficient information for accurate measurement determination, discarding redundant data while maintaining measurement precision
Solution Approach 2:
The patent uses partial waveform data (short patterns rather than complete long sequences) to achieve sufficient measurement accuracy. By analyzing only the critical portions of waveforms that contain the necessary information for measurement determination, the system achieves adequate precision without the computational burden of processing complete extended waveforms
3Productivity
If conventional measurement methods without time sequence information are used, then processing speed is faster, but measurement accuracy deteriorates due to inability to capture equalizer effects
Solution Approach 1:
The patent creates waveform pattern databases that copy and store the characteristic time sequence information of signals at different equalizer settings. During measurement, the system compares captured waveforms against these pre-stored patterns to rapidly determine measurements, achieving both speed through pattern matching and accuracy through preservation of time sequence characteristics in the database entries
Data Source
AI summary
A test and measurement system includes a test and measurement device configured to receive a signal from a device under test, and one or more processors configured to execute code that causes the one or more processors to generate a waveform from the signal, apply an equalizer to the waveform, receive an input identifying one or more measurements to be made on the waveform, select a number of unit intervals (UIs) for a known data pattern, scan the waveform for the known data patterns having a length of the number of UIs, identify the known data patterns as short pattern waveforms, apply a machine learning system to the short pattern waveforms to obtain a value for the one or more measurements, and provide the values of the one or more measurements for the waveform. A method includes receiving a signal from a device under test, generating a waveform from the signal, applying an equalizer to the waveform, receiving an input identifying one or more measurements to be made on the waveform, selecting a number of unit intervals (UIs), scanning the waveform to identify short pattern waveforms having a length equal to the number of UIs, applying a machine learning system to the short pattern waveforms to obtain a value for the one or more measurements, and providing the values of the one or more measurements for the waveform from the machine learning system.