|
|
|
| R2012a Documentation → Simulink Verification and Validation | |
Learn more about Simulink Verification and Validation |
|
|
|
|
| R2012a Documentation → Simulink Verification and Validation | |
Learn more about Simulink Verification and Validation |
|
| Contents | Index |
To generate a Model Coverage report, select Tools > Coverage Settings and specify the desired options on the Reporting tab of the Coverage Settings dialog box. For Stateflow charts, the Simulink Verification and Validation software records the execution of the chart itself and the execution of states, transition decisions, and individual conditions that compose each decision. After simulation ends, the model coverage reports on how thoroughly a model was tested. The report shows:
How many times each exclusive substate is entered, executed, and exited based on the history of the superstate
How many times each transition decision has been evaluated as true or false
How many times each condition has been evaluated as true or false
For complete information about Stateflow software, see Stateflow User's Guide.
To specify coverage report settings, select Tools > Coverage Settings in a Simulink model window.
By selecting the Generate HTML Report option in the Coverage Settings dialog box, you can create an HTML report containing the coverage data generated during simulation of the model. The report appears in the MATLAB Help browser at the end of simulation.
By selecting the Generate HTML Report option, you also enable the selection of different coverages that you can specify for your reports. The following sections address only coverage metrics that affect reports for Stateflow charts. These metrics include decision coverage, condition coverage, and MCDC coverage. For a complete discussion of all dialog box fields and entries, see Setting Model Coverage Options.
Cyclomatic complexity is a measure of the complexity of a software module based on its edges, nodes, and components within a control-flow graph. It provides an indication of how many times you need to test the module.
The calculation of cyclomatic complexity is as follows:
CC = E - N + p
where CC is the cyclomatic complexity, E is the number of edges, N is the number of nodes, and p is the number of components.
Within the Model Coverage tool, each decision is exactly equivalent to a single control flow node, and each decision outcome is equivalent to a control flow edge. Any additional structure in the control-flow graph is ignored since it contributes the same number of nodes as edges and therefore has no effect on the complexity calculation. Therefore, you can express cyclomatic complexity as follows:
CC = OUTCOMES - DECISIONS + p
For analysis purposes, each chart counts as a single component.
Decision coverage interprets a model execution in terms of underlying decisions where behavior or execution must take one outcome from a set of mutually exclusive outcomes.
Note Full coverage for an object of decision means that every decision has had at least one occurrence of each of its possible outcomes. |
Decisions belong to an object making the decision based on its contents or properties. The following table lists the decisions recorded for model coverage for the Stateflow objects owning them. The sections that follow the table describe these decisions and their possible outcomes.
Object | Possible Decisions |
|---|---|
Chart | If a chart is a triggered Simulink block, it must decide whether or not to execute its block. If a chart contains exclusive (OR) substates, it must decide which of its states to execute. |
State | If a state is a superstate containing exclusive (OR) substates, it must decide which substate to execute. If a state has on event name actions (which might include temporal logic operators), the state must decide whether or not to execute the actions. |
Transition | If a transition is a conditional transition, it must decide whether or not to exit its active source state or junction and enter another state or junction. |
If the chart is a triggered block in a Simulink model, the decision to execute the block is tested. If the block is not triggered, there is no decision to execute the block, and the measurement of decision coverage is not applicable (NA).
If the chart contains exclusive (OR) substates, the decision on which substate to execute is tested. If the chart contains only parallel AND substates, this coverage measurement is not applicable (NA).
Since a chart is hierarchically processed from the top down, procedures such as exclusive (OR) substate entry, exit, and execution are sometimes decided by the parenting superstate.
Note Decision coverage for superstates applies only to exclusive (OR) substates. A superstate makes no decisions for parallel (AND) substates. |
Since a superstate must decide which exclusive (OR) substate to process, the number of decision outcomes for the superstate is the number of exclusive (OR) substates that it contains. In the examples that follow, the choice of which substate to process can occur in one of three possible contexts.
| Context | Example | Decisions That Occur |
|---|---|---|
States A and A1 are active.
|
During processing of state A1, all outgoing transitions are tested. This decision belongs to the transition and not to the parent state A. In this case, the transition marked by condition C2 is tested and a decision is made whether to take the transition to A2 or not. | |
A transition takes place whose source is superstate A and whose destination is state B.
| If the superstate has two exclusive (OR) substates, it is the decision of superstate A which substate performs the implicit transition from substate to superstate. | |
A history junction records which substate was last active before the superstate was exited.
| If that superstate becomes the destination of one or more transitions, the history junction decides which previously active substate to enter. |
For more information, see State Details Report Section.
A state that has an on event_name action statement must decide whether to execute that statement based on the reception of a specified event or on an accumulation of the specified event when using temporal logic operators.
A conditional transition is a transition with a triggering event and/or a guarding condition. In a conditional transition from one state to another, the decision to exit one state and enter another is credited to the transition itself.
Note Only conditional transitions receive decision coverage. Transitions without decisions are not applicable to decision coverage. |
Condition coverage reports on the extent to which all possible outcomes are achieved for individual subconditions composing a transition decision.
Note Full condition coverage means that all possible outcomes occurred for each subcondition in the test of a decision. |
For example, for the decision [A & B & C] on a transition, condition coverage reports on the true and false occurrences of each of the subconditions A, B, and C. This results in eight possible outcomes: true and false for each of three subconditions.
| Outcome | A | B | C |
|---|---|---|---|
| 1 | T | T | T |
| 2 | T | T | F |
| 3 | T | F | T |
| 4 | T | F | F |
| 5 | F | T | T |
| 6 | F | T | F |
| 7 | F | F | T |
| 8 | F | F | F |
For more information, see Transition Details Report Section.
The Modified Condition Decision Coverage (MCDC) option reports a test's coverage of occurrences in which changing an individual subcondition within a transition results in changing the entire transition trigger expression from true to false or false to true.
For example, if a transition executes on the condition [C1 & C2 & C3 | C4 & C5], the MCDC report for that transition shows actual occurrences for each of the five subconditions (C1, C2, C3, C4, C5) in which changing its result from true to false is able to change the result of the entire condition from true to false.
You can use the following Simulink Design Verifier functions inside Stateflow charts:
If you do not have a Simulink Design Verifier license, you can collect model coverage for a Stateflow chart containing these functions, but you cannot analyze the model using the Simulink Design Verifier software.
When you specify the Simulink Design Verifier coverage metric in the Coverage Settings dialog box, the Simulink Verification and Validation software records coverage for these functions.
Each of these functions evaluates an expression expr, for example, sldv.test(expr), where expr is any valid Boolean MATLAB expression. Simulink Design Verifier coverage measures the number of time steps that the expression expr evaluates to true.
If expr is true for at least one time step, Simulink Design Verifier coverage for that function is 100%. Otherwise, the Simulink Verification and Validation software reports coverage for that function as 0%.
Consider a model that contains this Stateflow chart:
To collect coverage for Simulink Design Verifier functions, on the Coverage Settings dialog box, on the Coverage tab, select Simulink Design Verifier.
After simulation, the model coverage report lists coverage for the sldv.condition, sldv.assume, sldv.prove, and sldv.test functions.
The following sections of a Model Coverage report were generated by simulating the sf_boiler model, which includes the Bang-Bang Controller chart. The coverage metrics for Decision, Condition, and MCDC are enabled for this report.
The Summary section shows coverage results for the entire test and appears at the beginning of the Model Coverage report.
Each line in the hierarchy summarizes the coverage results at that level and the levels below it. You can click a hyperlink to a later section in the report with the same assigned hierarchical order number that details that coverage and the coverage of its children.
The top level, sf_boiler, is the Simulink model itself. The second level, Bang-Bang Controller, is the Stateflow chart. The next levels are superstates within the chart, in order of hierarchical containment. Each superstate uses an SF: prefix. The bottom level, Boiler Plant model, is an additional subsystem in the model.
When recording coverage for a Stateflow chart, the Simulink Verification and Validation software reports two types of coverage for the chart—Subsystem and Chart.
Subsystem — This section reports coverage for the chart:
Coverage (this object): Coverage data for the chart as a container object
Coverage (inc.) descendants: Coverage data for the chart and the states and transitions in the chart.
If you click the hyperlink of the subsystem name in the section title, the Bang-Bang Controller block is highlighted in the block diagram.
Decision coverage is not applicable (NA) because this chart does not have an explicit trigger. Condition coverage and MCDC are not applicable (NA) for a chart, but apply to its descendants.
Chart — This section reports coverage for the chart:
Coverage (this object): Coverage data for the chart and its inputs
Coverage (inc.) descendants: Coverage data for the chart and the states and transitions in the chart.
If you click the hyperlink of the chart name in the section title, the chart opens in the Stateflow Editor.
Decision coverage is listed appears for the chart and its descendants. Condition coverage and MCDC are not applicable (NA) for a chart, but apply to its descendants.
For each state in a chart, the coverage report includes a State section with details about the coverage recorded for that state.
In the sf_boiler model, the state On resides in the box Heater. On is a superstate that contains:
Two substates HIGH and NORM
A history junction
The function warm
The coverage report includes a State section on the state On.
The decision coverage for the On state tests the decision of which substate to execute.
The three decisions are listed in the report:
Under Substate executed, which substate to execute when On executes.
Under Substate exited when parent exited, which substate is active when On exits. NORM is listed as never being active when On exits because the coverage tool sees the supertransition from NORM to Off as a transition from On to Off.
Under Previously active substate entered due to history, which substate to reenter when On re-executes. The history junction records the previously active substate.
Because each decision can result in either HIGH or NORM, the total possible outcomes are 3 × 2 = 6. The results indicate that five of six possible outcomes were tested during simulation.
Cyclomatic complexity and decision coverage also apply to descendants of the On state. The decision required by the condition [warm()] for the transition from HIGH to NORM brings the total possible decision outcomes to 8. Condition coverage and MCDC are not applicable (NA) for a state.
Note Nodes and edges that make up the cyclomatic complexity calculation have no direct relationship with model objects (states, transitions, and so on). Instead, this calculation requires a graph representation of the equivalent control flow. |
Reports for transitions appear under the report sections of their owning objects. Transitions do not appear in the model hierarchy of the Summary section, since the hierarchy is based on superstates that own other Stateflow objects.
The decision for this transition depends on the time delay of 40 seconds and the condition [cold()]. If, after a 40 second delay, the environment is cold (cold() = 1), the decision to execute this transition and turn the Heater on is made. For other time intervals or environment conditions, the decision is made not to execute.
For decision coverage, both true and false outcomes occurred. Because two of two decision outcomes occurred, coverage was full or 100%.
Condition coverage shows that only 4 of 6 condition outcomes were tested. The temporal logic statement after(40,sec) represents two conditions: the occurrence of sec and the time delay after(40,sec). Therefore, three conditions on the transition exist: sec, after(40,sec), and cold(). Since each of these decisions can be true or false, six possible condition outcomes exist.
The Conditions analyzed table shows each condition as a row with the recorded number of occurrences for each outcome (true or false). Decision rows in which a possible outcome did not occur are shaded. For example, the first and the third rows did not record an occurrence of a false outcome.
In the MC/DC report, all sets of occurrences of the transition conditions are scanned for a particular pair of decisions for each condition in which the following are true:
The condition varies from true to false.
All other conditions contributing to the decision outcome remain constant.
The outcome of the decision varies from true to false, or the reverse.
For three conditions related by an implied AND operator, these criteria can be satisfied by the occurrence of these conditions.
Condition Tested | True Outcome | False Outcome |
|---|---|---|
1 | TTT | Fxx |
2 | TTT | TFx |
3 | TTT | TTF |
Notice that in each line, the condition tested changes from true to false while the other condition remains constant. Irrelevant contributors are coded with an "x" (discussed below). If both outcomes occur during testing, coverage is complete (100%) for the condition tested.
The preceding report example shows coverage only for condition 2. The false outcomes required for conditions 1 and 3 did not occur, and are indicated by parentheses for both conditions. Therefore, condition rows 1 and 3 are shaded. While condition 2 has been tested, conditions 1 and 3 have not and MCDC is 33%.
For some decisions, the values of some conditions are irrelevant under certain circumstances. For example, in the decision [C1 & C2 & C3 | C4 & C5] the left side of the | is false if any one of the conditions C1, C2, or C3 is false. The same applies to the right side result if either C4 or C5 is false. When searching for matching pairs that change the outcome of the decision by changing one condition, holding some of the remaining conditions constant is irrelevant. In these cases, the MC/DC report marks these conditions with an "x" to indicate their irrelevance as a contributor to the result. These conditions appear as shown.
Consider the first matched pair. Since condition 1 is true in the True outcome column, it must be false in the matching False outcome column. This makes the conditions C2 and C3 irrelevant for the false outcome since C1 & C2 & C3 is always false if C1 is false. Also, since the false outcome is required to evaluate to false, the evaluation of C4 & C5 must also be false. In this case, a match was found with C4 = F, making condition C5 irrelevant.
In a Stateflow chart, an atomic subchart is a graphical object that allows you to reuse the same state or subchart across multiple charts and models.
When you specify to record coverage data for a model during simulation, the Simulink Verification and Validation software records coverage for any atomic subcharts in your model. The coverage data records the execution of the chart itself, and the execution of states, transition decisions, and individual conditions that compose each decision in the atomic subchart.
Simulate the doc_atomic_subcharts_map_iodata example model and record decision coverage:
Open the doc_atomic_subcharts_map_iodata.mdl model.
This model contains two Sine Wave blocks that supply input signals to the Stateflow chart Chart. Chart contains two atomic subcharts—A and B—that are linked from the same library chart, also named A. The library chart contains the following objects:
In the Model Editor, select Tools > Coverage Settings
The Coverage Settings dialog box appears.
On the Coverage tab, select Coverage for this model: doc_atomic_subcharts_map_iodata.
On the Reporting tab, select Generate HTML report.
Click OK to close the Coverage Settings dialog box.
Simulate the doc_atomic_subcharts_map_iodata model.
When the simulation completes, the coverage report opens.
The report provides coverage data for atomic subcharts A and B in the following forms:
For the atomic subchart instance and its contents. Decision coverage is not applicable (NA) because this chart does not have an explicit trigger.
For the library chart A and its contents. The chart itself achieves 100% coverage on the input u1, and 88% coverage on the states and transitions inside the library chart.
Atomic subchart B is a copy of the same library chart A. The coverage of the contents of subchart B is identical to the coverage of the contents of subchart A.
Simulink Verification and Validation software reports model coverage for the decisions the objects make in a Stateflow chart during model simulation. The report includes coverage for the decisions the truth table functions make.
| For this type of truth table... | The report includes coverage data for... |
|---|---|
| Stateflow Classic | Conditions only. |
| MATLAB | Conditions and only those actions that have decision points. |
For detailed information about truth tables and truth table functions in Stateflow, see Truth Table Functions for Decision-Making Logic.
If you have a Stateflow license, you can generate a model coverage report for a truth table.
Consider the following model.
The Stateflow chart Chart contains the following truth table:
When you simulate the model and collect coverage, the model coverage report includes the following data:
The Coverage (this object) column shows no coverage. The reason is that the container object for the truth table function—the Stateflow chart—does not decide whether to execute the ttable truth table.
The Coverage (inc. descendants) column shows coverage for the graphical function. The graphical function has the decision logic that makes the transitions for the truth table. The transitions in the graphical function contain the decisions and conditions of the truth table. Coverage for the descendants in the Coverage (inc. descendants) column includes coverage for these conditions and decisions. Function calls to the truth table test the model coverage of these conditions and decisions.
Note See How Stateflow Software Generates Content for Truth Tables for a description of the graphical function for a truth table. |
Coverage for the decisions and their individual conditions in the ttable truth table function are as follows.
The Model Coverage tool displays model coverage results for individual blocks directly in Simulink diagrams. If you enable this feature, the Model Coverage tool:
Highlights Stateflow objects that receive model coverage during simulation
Provides a context-sensitive display of summary model coverage information for each object
For details on enabling and selecting this feature in the Simulink window, see Enabling Coverage Highlighting in the Simulink Verification and Validation documentation.
Once you enable display coverage with model coloring, anytime that the model generates a model coverage report, individual chart objects receiving coverage appear highlighted with light green or light red.
Open the sf_car model.
Select Tools > Coverage Settings.
In the Coverage Settings dialog box, select Coverage for this model.
Click OK.
Simulate the model.
After simulation ends, chart objects with coverage appear highlighted.
Object highlighting indicates coverage as follows:
Light green for full coverage
Light red for partial coverage
No color for zero coverage
Click selection_state in the chart.
The following summary report appears.
When you click a highlighted Stateflow object, the summarized coverage for that object appears in the Coverage Display Window. Clicking the hyperlink opens the section of the coverage report for this object.
Tip You can set the Coverage Display Window to appear for a block in response to a hovering mouse cursor instead of a mouse click in one of two ways:
|
 | Model Coverage for MATLAB Functions | Results Review |  |
Learn more about Simulink through this collection of videos, articles, technical literature and the Getting Started with Simulink Guide.
| © 1984-2012- The MathWorks, Inc. - Site Help - Patents - Trademarks - Privacy Policy - Preventing Piracy - RSS |
