Distinguishing Characteristics
Trusted in High Assurace Domains
T-VEC has been applied to high assurance systems which have undergone the Federal Avionics Administration's certification process. The FAA accepted T-VEC artifacts as trusted outputs, such that the automatically generated test vectors, coverage reports, and results comparison reports did not have to undergo any manual review processes, resulting in dramatic savings in cost, and cycle time and increased product quality. T-VEC satisfies Modified Condition Decision Coverage (MCDC) in excess of DO-178B.
Expected Outputs Computed
Computation of expected outputs for each test vector, providing a unique solution to a key problem area in test generation known as the test oracle problem. The test oracle problem has resulted from automated tools that select test-input data but do not support verification of the actual test outputs. Determining the expected output for a set of test point inputs requires a specification that characterizes the desired system function for every functional relationship of a system. A specification model should be developed to level of detail that allows a test generation tool to produce expected output values as well as the test inputs.
Support for Complex and Floating-point Types
Support for complex aggregate data space, constraint, and computation definition.
Support for nonlinear inequalities
Support for solving problems with nonlinear inequalities as well as linear inequalities, where both sides of the inequality can be expressions with variables.
Availability of various test selection heuristics that extend the principles of domain testing theory to localize system faults. Highly effective heuristics strategically handle various data types, constraint solution area corner point selection, and combinatorial independence considerations. Specialized heuristics are available to provide additional fault detection capabilities.
Scales to Industrial Applications
Our mechanisms preclude the combinatorial explosion problem associated with tests generated from the constraint combinations in subsystem hierarchies (commonly found in integration testing).
Hierarchical Specification Model
The concept of hierarchical specification is fundamental to the scalability of both the specification method and the associated verification processes. T-VEC promotes a hierarchical specification to manage complexity, change, and reuse, as well as scalability.
Model Analysis Capabilities
Contradictions can be difficult to identify even in simple models, but with large hierarchical systems, model contradictions can be very difficult to identify. These conditions commonly indicate specification errors that with early detection can be corrected before propagating to other development artifacts. Inherent in our approach is the ability to detect unsatisfiable constraints in hierarchical specifications.
|
