Pattern Sensitive Algorithm Test Generation Using Feature Clustering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for testing pattern sensitive algorithms in semiconductor design, such as optical proximity correction, face challenges in identifying and addressing systematic defects in high-dimensional spaces, where direct assessment of coverage is impractical due to the large number of possible shape interactions.
Innovation Solution
A system and method that generates test patterns by extracting feature samples from layout designs, grouping them into clusters, and selecting areas that cover each cluster, ensuring comprehensive coverage while minimizing computational time, using techniques like Walsh patterns and clustering to identify representative regions for testing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If direct assessment of coverage by generating histograms of the space is used, then measurement precision of coverage is improved, but computational time and resource requirements increase significantly due to the large number of table entries required (kd where k is the number of bins and d is the number of feature dimensions)
Solution Approach 1:
The patent extracts only the essential feature samples from the layout design rather than processing the entire design space. By selecting representative samples that capture the critical shape interactions, the system achieves coverage assessment without requiring exhaustive histogram generation across all possible dimensions, thus reducing computational time while maintaining measurement precision.
Solution Approach 2:
The patent segments the high-dimensional feature space into discrete clusters of similar layout patterns. By organizing samples into clusters based on their feature characteristics, the system can assess coverage within each cluster separately, reducing the overall computational complexity from processing the entire high-dimensional space to processing manageable cluster representations.
2Reliability
If a comprehensive set of test cases is generated to maximize coverage of possible shape interactions, then reliability of defect identification is improved, but device complexity and difficulty of managing test cases increase
Solution Approach 1:
The patent extracts and selects only the most representative feature samples that are most likely to reveal systematic defects. By using clustering to identify and select samples from each cluster, the system generates a reduced but highly reliable test case set that maintains defect detection capability while significantly reducing the total number of test cases that need to be managed.
Solution Approach 2:
The patent performs preliminary clustering and selection of representative samples before generating the final test case set. This preliminary organization of samples into clusters and identification of representative members allows for systematic test case generation that is easier to manage and less complex than generating all possible combinations.
Data Source
AI summary
A system and method for generating test patterns for a pattern sensitive algorithm. The method comprises the steps extracting feature samples from a layout design; grouping feature samples into clusters; selecting at least one area from the layout design that covers a feature sample from each cluster; and saving each pattern layout covered by the at least one area as test patterns.