Handling Incompatibilities with Automatic Stubbing :: How the Simulink Design Verifier Software Works (Simulink® Design Verifier™)

When you enable the automatic stubbing option (it is enabled by default), the software considers only the interface of the unsupported objects, not their actual behavior. This technique allows the software to complete the analysis. However, the analysis may achieve only partial results if any unsupported model element affects the simulation outcome.

How Automatic Stubbing Works

If you enable automatic stubbing, when the Simulink Design Verifier analysis comes to an unsupported block, the software "stubs" that block. The analysis ignores the behavior of the block, and as a result, the block output can take any value.

Stubbing Example: Trigonometric Function Block

The Simulink Design Verifier software does not support Trigonometric Function blocks when the Function parameter is set to acos, such as the one in the following graphic.

When stubbing this block during analysis, out_signal can take any value, with the following results.

Analysis Model Result of Stubbing out_signal

Analysis Model	Result of Stubbing out_signal
Design error detection	If a design-error objective that depends on `out_signal` is proven valid, that objective is valid for all simulations. In this case, the stubbing did not affect the results of the analysis. If a design-error objective that depends on `out_signal` is falsified, the analysis cannot create a test case. The analysis cannot determine which input to the stubbed block produces the output that falsifies the objective.
Test case generation	If a test objective that depends on the value of `out_signal` is satisfied, the analysis cannot create a test case. The analysis cannot determine which input to the stubbed block produces the output that satisfies the objective. If a test objective that depends on the value of `out_signal` is unsatisfiable, there is no simulation that can satisfy that objective. In this case, the stubbing did not affect the results of the analysis.
Property proving	If a proof objective that depends on `out_signal` is proven valid, that objective is valid for all simulations. In this case, the stubbing did not affect the results of the analysis. If a proof objective that depends on `out_signal` is falsified, the analysis cannot create a counterexample. The analysis cannot determine which input to the stubbed block produces the output that falsifies the objective.

Design error detection

If a design-error objective that depends on out_signal is proven valid, that objective is valid for all simulations. In this case, the stubbing did not affect the results of the analysis.
If a design-error objective that depends on out_signal is falsified, the analysis cannot create a test case. The analysis cannot determine which input to the stubbed block produces the output that falsifies the objective.

Test case generation

If a test objective that depends on the value of out_signal is satisfied, the analysis cannot create a test case. The analysis cannot determine which input to the stubbed block produces the output that satisfies the objective.
If a test objective that depends on the value of out_signal is unsatisfiable, there is no simulation that can satisfy that objective. In this case, the stubbing did not affect the results of the analysis.

Property proving

If a proof objective that depends on out_signal is proven valid, that objective is valid for all simulations. In this case, the stubbing did not affect the results of the analysis.
If a proof objective that depends on out_signal is falsified, the analysis cannot create a counterexample. The analysis cannot determine which input to the stubbed block produces the output that falsifies the objective.

Stubbing Example: S-Function Blocks and Function-Call Triggers

The Simulink demo model sfcndemo_sfun_fcncall has an S-Function block. The S-function sfun_fcncall triggers the execution of the function-call subsystems f1 subsys1 and f2 subsys2 on the first and second elements of the first output port.

The Simulink Design Verifier software does not support the S-Function block, so if automatic stubbing is enabled, the analysis ignores the behavior of the S-function. As a result, the code that triggers the two function-call subsystems is ignored, resulting in two unsatisfiable objectives. Since the function calls are ignored, the contents of those subsystems are effectively eliminated from the analysis.

Analyzing a Model Using Automatic Stubbing

This section describes a workflow for using automatic stubbing, with a simple Simulink model as an example.

The following model contains a Trigonometric Function block that is not compatible with the Simulink Design Verifier software.

Checking Model Compatibility

From the Model Editor, there are two ways to check whether a model is compatible with the Simulink Design Verifier software:

Run the Simulink Design Verifier compatibility check by selecting Tools > Design Verifier > Check Model Compatibility.
Select the analysis that you want:
- Tools > Design Verifier > Detect Design Errors
- Tools > Design Verifier > Generate Tests
- Tools > Design Verifier > Prove Properties
The software first checks the compatibility of the model. If the model itself is incompatible, for example, if it uses a variable-step solver, the analysis cannot continue.
If it finds incompatible elements in the model, the software analyzes the model, and by default, stubs out the incompatible elements. The Simulation Diagnostics Viewer also opens, listing the incompatibilities.

Turning On Automatic Stubbing

Automatic stubbing is enabled by default. To change the automatic stubbing setting, in the Configuration Parameters dialog box, on the main Design Verifier pane, select Automatic stubbing of unsupported block and functions. When you run the analysis, the software tells you that stubbing is turned on and the analysis continues.

Reviewing the Results

If you run an analysis with automatic stubbing enabled, make sure to review the results. In this report, generated after a test case generation analysis, you see a table of unsupported blocks that the software encountered.

The Summary report for the example model shows that one objective was satisfied without generating a test case. The software cannot generate the test case because it does not understand the operation of the Trigonometric Function block.

Achieving Complete Results

If your analysis does not achieve complete results because of the stubbing, you can define custom block replacements to give a more precise definition of the unsupported blocks. For more information:

See
Defining Custom Block Replacements.
At the MATLAB command line, enter
```
echodemo sldvdemo_blockreplacement_unsupportedblocks 
```
to step through the "Block Replacements for Unsupported Blocks" demo.

Analyzing Large Models

Handling Nonfinite Data

Related Products & Applications

Get Simulink trial software

Learn more about Simulink through this collection of videos, articles, technical literature and the Getting Started with Simulink Guide.

Get this Simulink Kit

Search MATLAB Documentation
R2012a Documentation → Simulink Design Verifier
View documentation for other releases	Learn more about Simulink Design Verifier