Event Correlation Detection for Hardware Verification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current verification methods for hardware and software, such as those using logic analyzers or tracers, face inefficiencies due to large output sizes and difficulties in analyzing event correlations, particularly when dealing with long-term operations, and often require extensive memory and preliminary modeling of statistical distributions.
Innovation Solution
A detecting apparatus and method that calculates the degree of correlation between first and second events by acquiring and measuring event count values over specific periods, allowing for efficient verification and debugging without the need for preliminary registration of expected event patterns or extensive memory usage.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If verification methods using logic analyzers or tracers are applied to large hardware or software systems, then verification coverage is improved, but output data volume increases enormously requiring sequential filtering
Solution Approach 1:
The patent extracts only the essential verification information by calculating correlation coefficients between event sequences, rather than outputting all raw event data. This allows verification coverage to be maintained while dramatically reducing the volume of output data that requires analysis.
Solution Approach 2:
Instead of sequentially filtering output results to find relevant bugs, the patent inverts the approach by pre-calculating event correlations and using these correlations to guide verification, thereby avoiding the need to process enormous amounts of raw output data.
2Measurement precision
If statistical verification methods are used to analyze event correlations, then verification accuracy is improved, but memory requirements increase for long-term operations
Solution Approach 1:
The patent segments the event analysis process into discrete correlation calculations between specific event sequences, rather than attempting to analyze all events simultaneously. This allows accurate correlation measurement while using minimal memory by processing events in manageable segments.
Solution Approach 2:
The patent changes the parameter representation from storing complete event sequences to storing compact correlation coefficients and event timing data. This transformation maintains verification accuracy while dramatically reducing memory requirements for long-term operational data.
3Measurement precision
If preliminary modeling of statistical distributions is required for verification, then detection accuracy for abnormal events is improved, but device complexity and preliminary setup time increase
Solution Approach 1:
The patent enables the verification system to automatically calculate event correlations and identify abnormal patterns without requiring external preliminary modeling. The system serves itself by deriving verification criteria directly from the observed event sequences, eliminating the need for complex pre-configured statistical models.
Data Source
Figure 1
Figure 2
Figure 3~4
AI summary
[Object] To calculate in a simple manner the degree of correlation between first events and second events repeatedly occurring in an observed apparatus. [Solution] There is provided a detecting apparatus that detects the degree of correlation between first events and second events repeatedly occurring in an observed apparatus, comprising: an acquiring unit that acquires second event count values each indicating the number of second events occurring during each first period between each first event and the first event next thereto; a measuring unit that measures an observed number of each second event count value derived from the number of times the second event count value is observed; and a calculating unit that calculates the degree of correlation between the first events and the second events based on the observed number of each second event count value.