Symbolic Test Generation for Reactive Systems

1
Symbolic Test Generation for Reactive Systems

• Elena Zinovieva
• IRISA / INRIA Rennes
• 24 October, 2001

2
Testing
Is there difference between expected and actual
behavior of the system?
Test Failed (there are problems with the system
implementation)
Test Passed
3
White-Box Testing
White-Box Testing is based on the internal
structure of a computer system. The aim of such
a testing is to exercise thoroughly the systems
code.
4
Black-Box Testing
In Black-Box Testing the computer system is
treated as a black-box, no references are made to
the internal structure, so it is possible to
observe only inputs/outputs to/from the system.
5
Kinds of the Testing

• White-Box Testing
• Data-Flow Testing
• Mutation Testing
• Domain Testing

• Black-Box Testing
• Performance Testing
• Robustness Testing
• Reliability Testing
• Conformance Testing

6
Example Tea Machine
tea button
coffee
tea
7
Reactive System
8
Software Development
Specification, Test Purpose
System Requirements
IUT
Test Failed an error is in the IUT
Executable Test Case

• Test Case Generation
• TGV
• TorX
• TestComposer
• STG.

Test Passed the system works fine
Test Execution
9
Problems and Solution
Possible

• State-space explosion as variables are
  instantiated with all their possible values
• Test cases are not readily understandable.

Use a new model and symbolic techniques to
produce a test case!
10
IOSTS Model (2/1)
Specification
11
IOSTS Model (2/2)
Test Purpose
12
STG Architecture
Symbolic Test Case IOSTS
Symbolic Specification IOSTS
Symbolic Test Generation
Symbolic Test Purpose IOSTS
Translation IOSTS to C
Executable Test Program C
13
Symbolic Test Generation
14
Test Case
(vPaidgtpPrice) (mRemValvPaid-pPrice) Return ?
(mRemVal) vPaidpPrice
Test Case
15
Results

• The derivation of test cases from formal
  specifications can be fully automated.
• The test cases are symbolic and concrete.
• The test cases are correct (no false positive
  outcomes, no false negative outcomes, etc.)

16
Ongoing and Future Work

• Improving heuristic for the simplification
  (symbolic test generation) using a constraint
  propagation.
• Use higher-level languages (LOTOS, etc.) for
  specifying systems at the user level, which
  translates automatically to IOSTS.
• Work on the implementation of mechanism to
  automatically compute test purpose from the
  system specification.