Main Content

Perform Operations on Coverage Data

This example shows how to use the overloaded operators +, *, and - to combine coverage results into a union, intersection, or set difference of results.

Open Model

Open the slvnvdemo_cv_mutual_exclusion model. The model contains a Constant block connected to the enable port of two subsystems. One of the subsystems has a NOT block placed before the enable port, which means only one of the subsystems can be enabled at a time.

modelName = "slvnvdemo_cv_mutual_exclusion";
open_system(modelName)

Create a Simulink.SimulationInput object and then use setModelParameter to set some coverage parameters.

simIn = Simulink.SimulationInput(modelName);

Enable coverage analyze for the model by setting the CovEnable parameter to on.

simIn = setModelParameter(simIn,"CovEnable","on");

Set the structural coverage level to Decision.

simIn = setModelParameter(simIn,"CovMetricStructuralLevel","Decision");

Display the coverage data in MATLAB® instead of the file location by setting CovSaveOutputData to off. This also prevents Simulink® Coverage® from automatically saving the coverage data to a file after the simulation.

simIn = setModelParameter(simIn,"CovSaveOutputData","off");

Simulate the model by using simIn as the input to sim. The SimulationOutput object contains the coverage data as a property whose name is equal to the value of the CovSaveName configuration parameter. The default value is covdata.

simOut1 = sim(simIn);
data1 = simOut1.covdata
data1 = ... cvdata
            version: (R2024b)
                 id: 237
               type: TEST_DATA
               test: cvtest object
             rootID: 239
           checksum: [1x1 struct]
          modelinfo: [1x1 struct]
          startTime: 20-Jul-2024 19:13:42
           stopTime: 20-Jul-2024 19:13:45
  intervalStartTime: 
   intervalStopTime: 
simulationStartTime: 0
 simulationStopTime: 10
             filter: 
            simMode: Normal

Change which subsystem is enabled by using the setBlockParameter method to change the value of the Constant block from 0 to 1, and then simulate the model again.

simIn = setBlockParameter(simIn,modelName+"/Constant","value","1");
simOut2 = sim(simIn);
data2 = simOut2.covdata
data2 = ... cvdata
            version: (R2024b)
                 id: 290
               type: TEST_DATA
               test: cvtest object
             rootID: 239
           checksum: [1x1 struct]
          modelinfo: [1x1 struct]
          startTime: 20-Jul-2024 19:13:47
           stopTime: 20-Jul-2024 19:13:47
  intervalStartTime: 
   intervalStopTime: 
simulationStartTime: 0
 simulationStopTime: 10
             filter: 
            simMode: Normal

Use the decisioninfo function to extract the decision coverage from each simulation and determine the percentage of decision outcomes satisfied.

cov1 = decisioninfo(data1,modelName);
percent1 = 100*(cov1(1)/cov1(2))

cov2 = decisioninfo(data2,modelName);
percent2 = 100*(cov2(1)/cov2(2))
percent1 =

    50


percent2 =

    50

Both simulations have 50% coverage. To check if the two simulations cover the same 50% of decision outcomes, look at the union and intersection of the two objects.

Find the Union of Coverage

Use the + operator to derive a third cvdata object that represents the union of data1 and data2 cvdata objects. The union of two or more cvdata objects is also referred to as cumulative coverage or aggregated coverage.

When you create cvdata objects by combining other simulation results, the type property of the new object is DERIVED_DATA.

dataUnion = data1 + data2
dataUnion = ... cvdata
            version: (R2024b)
                 id: 0
               type: DERIVED_DATA
               test: []
             rootID: 239
           checksum: [1x1 struct]
          modelinfo: [1x1 struct]
          startTime: 20-Jul-2024 19:13:42
           stopTime: 20-Jul-2024 19:13:47
  intervalStartTime: 
   intervalStopTime: 
             filter: 
            simMode: Normal

Extract the decision coverage results and determine the percentage of decision outcomes satisfied by calling the decisioninfo function on dataUnion.

covU = decisioninfo(dataUnion,modelName);
percentU = 100*(covU(1)/covU(2))
percentU =

   100

The union of the two simulations reports that 100% decision outcomes are satisfied. This result indicates that there is no overlap in the coverage between the two simulations and that all decision outcomes are satisfied.

Find the Intersection of Coverage

Confirm that the coverage does not overlap between the two simulations by intersecting data1 and data2 with the * operator. The intersection returns only the coverage outcomes that are satisfied in both cvdata objects.

dataIntersection = data1 * data2

covI = decisioninfo(dataIntersection,modelName);
percentI = 100*(covI(1)/covI(2))
dataIntersection = ... cvdata
            version: (R2024b)
                 id: 0
               type: DERIVED_DATA
               test: []
             rootID: 239
           checksum: [1x1 struct]
          modelinfo: [1x1 struct]
          startTime: 20-Jul-2024 19:13:42
           stopTime: 20-Jul-2024 19:13:47
  intervalStartTime: 
   intervalStopTime: 
             filter: 
            simMode: Normal


percentI =

     0

There is 0% decision coverage in the intersection because there is no overlap in coverage between the two simulations.

Compute the Coverage Difference

Use the - operator to create a cvdata object that represents the set difference between the left and right operands. The result of the operation contains the coverage outcomes that are satisfied in the left operand but not satisfied in the right operand. Use this operation to determine how much additional coverage is attributed to a particular simulation.

In this example, the difference between the union of the first and second simulation coverage and the first simulation coverage indicates how much additional coverage the second simulation provided. Because none of the decision coverage outcomes overlapped, the new decision coverage from the second simulation is 50%.

newCov2 = dataUnion - data1

covN = decisioninfo(newCov2,'slvnvdemo_cv_mutual_exclusion');
percentN = 100*(covN(1)/covN(2))
newCov2 = ... cvdata
            version: (R2024b)
                 id: 0
               type: DERIVED_DATA
               test: []
             rootID: 239
           checksum: [1x1 struct]
          modelinfo: [1x1 struct]
          startTime: 20-Jul-2024 19:13:42
           stopTime: 20-Jul-2024 19:13:47
  intervalStartTime: 
   intervalStopTime: 
             filter: 
            simMode: Normal


percentN =

    50

Use Derived Coverage Data Objects

You can use derived cvdata objects in all reporting and analysis functions and as inputs to subsequent operations. For example, generate a coverage report from the derived dataIntersection object and create a new cvdata union.

cvhtml('intersect_cov', dataIntersection);

newUnion = dataUnion + dataIntersection
newUnion = ... cvdata
            version: (R2024b)
                 id: 0
               type: DERIVED_DATA
               test: []
             rootID: 239
           checksum: [1x1 struct]
          modelinfo: [1x1 struct]
          startTime: 20-Jul-2024 19:13:42
           stopTime: 20-Jul-2024 19:13:47
  intervalStartTime: 
   intervalStopTime: 
             filter: 
            simMode: Normal

See Also

| |

Related Topics