Automated System-Level Test Generation via Segmented CSP Solving
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Automating the generation of system-level tests is hindered by the high computational cost and time required to create and execute simulated tests with numerous test vectors, especially for complex designs with many parameters, making it impractical and resource-intensive.
Innovation Solution
A method and system for automatically generating executable system-level tests by solving Constraint Satisfaction Problems (CSPs) for both the logic and data layers, scheduling actions and data paths, and assigning values to attributes, allowing for the creation of executable code that verifies system designs efficiently.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If simulated tests with numerous test vectors are used to verify complex system designs, then measurement precision and reliability are improved, but computing power consumption and time requirements increase enormously
Solution Approach 1:
The patent segments the test generation process into two distinct constraint satisfaction problems: a logic layer CSP that schedules actions and data paths, and a data layer CSP that assigns data attribute values. This segmentation allows each layer to be solved independently and sequentially, reducing the overall computational complexity compared to solving a single monolithic CSP with all constraints.
Solution Approach 2:
The patent performs preliminary scheduling of actions and data paths in the logic layer CSP before assigning specific data values in the data layer CSP. By establishing the logical structure and timing relationships first, the subsequent data assignment becomes a simpler constraint satisfaction problem with fewer variables and constraints to resolve simultaneously.
2Productivity
If automated generation of system-level test data models is performed, then productivity is improved, but computing power consumption increases prohibitively
Solution Approach 1:
The automated test generation process is divided into two sequential phases: logic layer processing that determines action scheduling and data path assignments, followed by data layer processing that assigns specific data values. This segmentation enables the system to generate tests automatically while consuming reasonable computing resources by avoiding the simultaneous resolution of all constraints in a single large CSP.
Solution Approach 2:
The system performs preliminary analysis and scheduling in the logic layer before proceeding to data assignment. By pre-determining the structural aspects of the test scenario including action sequences and data flow paths, the subsequent data generation phase becomes more efficient and requires less computational power.
3Ease of operation
If comprehensive test scenarios with complete action scheduling and data assignment are generated automatically, then ease of operation is improved, but device complexity increases
Solution Approach 1:
The complex task of automated test scenario generation is divided into two manageable constraint satisfaction problems with distinct responsibilities. The logic layer CSP handles structural concerns (action scheduling, data path selection) while the data layer CSP handles value assignment. This segmentation makes the overall system easier to operate and control despite the inherent complexity of automated generation.
Solution Approach 2:
The logic layer CSP acts as an intermediary between the user's test requirements and the detailed data layer CSP. By resolving the structural and scheduling constraints first, it creates a simplified framework that guides the subsequent data assignment process, making the overall complex system more manageable and easier to operate.
Data Source
AI summary
Method and system for automatically generating executable system-level tests. The method includes obtaining a system design including interrelation between components of the system design, actions the components are operable to perform, and constraints relating to the performance of the actions; receiving at least an initial action input to be tested; automatically generating a complete test scenario including: solving a logic layer CSP, including automatically scheduling actions and data paths of the test scenario, and assigning values to logic attributes of the actions and data, satisfying constraints relating to the logic layer, and solving a data layer CSP, satisfying constraints relating to data attributes taking into account the constraints relating to the logic layer; and generating the executable system-level test by assembling the initial action and the set of scheduled actions and data paths and the data attributes.