|
This section contains benchmarks of the classification 'test generation
benchmarks' and corresponding run-times of the RT-Tester tool whose model-based
test generation component RTT-MBT has been used as reference tool for creating
the tests.
Overview
Model description
The complete model description is available here.
Mutants description
Test cases description
File testcases specifies all test
cases related to model coverage criteria: each model coverage goal is
identified by means of a test case tag and a logical formula
specifying the constraint for covering the goal. The current baseline
of the benchmark model specifies the goals for
-
Basic Control State (BCS) coverage
The goals consist in reaching basic control states in
(potentially hierarchic) state machines of the model.
-
State Machine Transition (TR/HITR) coverage
The goals consist in covering transitions between (higher- or lower-level) control states of state machines.
-
MC/DC Coverage
The objective is again to cover state machine transitions,
with the refinement that guard conditions a && b should be
covered with different cases a && b, a && !b, !a && b, and guard
conditions a || b should be covered with test cases a && !b, !a
&& b, !a && !b, as far as possible (for more complex Boolean
expressions the MC/DC criteria are refined in a natural way).
-
User defined
The goals are expressed as LTL formula and provided by the user
For a description of all the configuration file please read
this.
Test Benchmarks
The benchmarks indicates also if the tests are able to kill the
mutants presented below . The mark ✔ means
that the test can kill the mutant, the mark ✘ is depicted otherwise.
-
The following two tests generates 100% BCS coverage.
No backtracking or random simulation techniques were
applied, only the SMT solver accelerated by abstract
interpretation.
The following tables refer to the
configuration of each test generation process and give short descriptions
of what the generated tests do.
-
The following tests generate 100% state machine transition
coverage. Two deal with only CSM_ON transitions, whereas
the other two deal with all the transition in the
model including high-level transition. The NOSB indicates that
the test cases consider that the train does not have service
brake.
State Machine MC/DC Coverage Tests
-
The following tests generate 100% state machine transition
coverage. The test procedure with NOSB assumes that no service
brake are available.
Requirement based testing
-
The following tests are requirement based testing. Requirements
are link to the model artifacts, transitions, state or constraints
(written as LTL formula). The test procedurws are generated
from a set of requirements, the tool provide the corresponding
tests that ensure that those requirement will be covered. The
requirements details may be found here.
-
The following two tests have been written by hand following the
definition given in [PH2013]. The fault domain considered here is D(S,m=6, I_2= I).
IECP (Input Equivalence Class Partioning) based testing (auto generated test suite)
-
The following tests were generated by a prototypical implementation of the
IECP test generation as described in [PH2013].
This implementation is not able
to deal with internal model variables and output variables that have an infinite
domain (e.g. floating point variables). Therefore a reduced version of the Ceiling Speed
Monitor model has been used.
This reduced model does only contain the lower-level state machine CSM_ON modelling the behaviour of the active
CSM and the only signals, used in this reduced
model, are V_est, V_max as inputs and DMICmd, DMIDisplaySBI, TICmd as
outputs. Furthermore it is assumed, that the service brake is available (SBAvailable = 1) and that, due to national regulations, the
emergency brakes must only be released after the train has come to a standstill (allowRevokeEB = 0).
The following tables show two exemplary test procedures that are part of the auto generated
IECP test suite. These test procedures can also be found in the zip file linked above, and are
just referenced here to give more details.
[PH2013]<> Jan Peleska and Wen-ling Huang:
Exhaustive Model-Based Equivalence Class Testing. In
Yenigün, Hüsnü and Yilmaz, Cemal and Ulrich, Andreas (eds.): Testing Software and Systems, Proceedings of the ICTSS2013. Springer, LNCS 8254, pp.49-64, 2013.
|