EXTRACTING THE MINIMIZED TEST SUITE FOR REVISED SIMULINK/STATEFLOW MODEL

International Journal of Artificial Intelligence and Applications (IJAIA), Vol.8, No.4, July 2017
DOI : 10.5121/ijaia.2017.8404 41
EXTRACTING THE MINIMIZED TEST SUITE FOR
REVISED SIMULINK/STATEFLOW MODEL
Han Gon Park1, Geon Gu Park1, Kihyun Chung1 and Kyunghee Choi2
1
Department of Electronic Engineering, Ajou University, Suwon, Republic of Korea
2
Department of Computer Engineering, Ajou University, Suwon, Republic of Korea
ABSTRACT
Test case generation techniques are successfully employed to generate test cases from a formal model. A
problem is that as the model evolves, test suites tend to grow in size, making it too costly to execute entire
test suites. This paper aims to propose a practical approach to reduce the size of test suites for modified
Simulink/Stateflow (SL/SF) model, which is popularly used for modeling software behavior in many
industries like automobile manufacturers. The model for describing a system is frequently modified until it
is fixed. The proposed technique is capable of extracting the minimized sized test suite in terms of test
coverage, by taking into account both the modified and the affected portion of revised SL/SF model. Two
real models for the ECUs deployed in a commercial car are used for an empirical study.
KEYWORDS
Test Suite Reduction, Simulink/Stateflow Model, Test Case Generation, Model Based Testing
1. INTRODUCTION
Software grows with time. The specification is modified for uses beyond the previous one, the
number of functions increases, the lines of code become longer, and many other complications
come with the growth of software. Some old portions are modified, some obsolete portions are
deleted, and some new portions are added. Due to such inevitable maintenance activities, more
than 70% of the time spent on software maintenance is in modifying and retesting the software.
To reduce the cost introduced by testing the revised software, efficient testing techniques are
definitely needed. The activity to revalidate the modified software is called regression testing.
Reference [1] identifies regression testing as being of two types: progressive regression testing,
and corrective regression testing. Progressive regression testing involves a modified specification,
while corrective regression testing does not perform specification change, but specification
correction. Most techniques focus on progressive testing, as do we in this paper.
Regression testing aims to ensure that no new faults are introduced into the previously validated
software due to the modification. The regression testing also needs to confirm that the newly
added requirements are properly integrated into the revised software. This is highly expensive,
time consuming, and laborious work. One main factor that makes regression testing expensive is
having to execute the whole set of test cases generated by the test generation method. A research
focus of regression testing techniques is to extract an effectively minimized sub set of test cases
from the whole test suite.
The test suite minimization problem is to select a subset of the entire test suite that can detect all
the faults. For the minimization, the test minimization techniques not only remove the test cases
from a suite that have become redundant with respect to a particular criterion or system output,
but they also need to add test cases that validate the newly added features. Coverage techniques

42
[2] select test cases that cause a modified program to satisfya criterion (depending on the
coverage criterion) that is different from that of the original program, while safe techniques [3]
select test cases to produce different output.
Since model-based regression testing of software has several advantages [4], our work is to
present a Simulink/Stateflow (SL/SF) model-based regression testing technique. We propose a
test case minimization for modified programs, of which the test cases are generated from an
SL/SFmodel [5].SL/SF has been popularly utilized in model-based development in many
industries, such as automobile manufacturers. They use SL/SF for modeling system behaviors,
describing specifications, producing auto codes [6], generating test cases [5], and utilizing the
model for testing systems [7].Extracting the minimized test suite from generated test casesbased
on an SL/SF model for the modified program is the focus of this study. The proposed technique
belongs to coverage technique.
Many previous coverage based test case selection techniques focus on removing test cases from a
test suite in such a way that redundant test cases are eliminated. The eliminated test suite should
hopefully contain the test cases for all the obsolete requirements, and the portions affected by the
obsolete requirements. If the specification is changed with some additional requirements, the
model has to be modified to include the additional requirements, and the test suite generated from
the revised model includes requirements for the added parts. In addition to the new test cases, the
modified software also needs to be tested by the test cases for the portions affected by the added
requirements.
To find the test cases for obsolete, added, and affected portions of a SL/SF model, we simulate
the old model and the revised model with test suites generated from the models. In simulating
SL/SF, it is possible to know whether each transition is evaluated as TRUE or FALSE, and which
state is active. Most test items consist of transition and state evaluation information in SL/SF
models. Two test suites generated from the original and modified models are applied to each
model, and every test case that covers a specific test item (for example, a state or transition
condition) of a model, but is not utilized to cover the test target of the other model, belongs to the
minimized test suite. The minimized test suite consists of the test cases found during the
simulation.
To confirm the effectiveness of the proposed technique, we conducted an experiment. As a
system-under-test, we used models for the Intelligent Management Module (IMS) and the Drive
Door Module (DDM) of a commercial passenger car provided by a major world class automobile
manufacturer. The result demonstrates that the proposed technique can be used to find the
minimized test suite for a revised SL/SF model that would otherwise be tested by the entire test
suite.
In the remainder of the paper, Section2 reviews the relevant literature, while Section 3 addresses
the proposed technique in detail. Section 4 describes the experiment results, while the Conclusion
summarizes the research.
2.RELATED WORK
Many approaches relating to regression test suite minimization have been published. Researchers
categorized regression test minimization techniques in different ways. One possible
categorization is test case minimization, test suite prioritization, and test suite selection.
The survey in Ref. [8] contained various minimization, selection, and prioritization techniques,
and discussed their open problems. Reference [9]used a combination of static slicing and delta
debugging that automatically minimizes the sequence of failure-inducing method calls, and

43
showed in a case study that the strategy could minimize failing unit test cases by 96%on average.
Reference [10] used additional criterion to break ties in the minimization process if there was
more than one test case with equal importance to a suite.
Reference [11] proposed a new metric to assess the rate of fault detection of prioritized test cases
that incorporate varying test cases and fault costs; they also presented the results of a case study
illustrating the application of the metric. Reference [12] presented results from an empirical study
of the application of several greedy, meta-heuristic, and evolutionary search algorithms that
belong to the prioritization technique category.
Reference [13] presented an approach to regression testing that handles the selection of test cases
from the existing test suite that should be rerun, and identification of the portions of the code that
must be covered by test cases. Both tasks were performed by traversing graphs for the program
and its modified version. Rothermel etal.[14] suggested a regression test selection technique for
use with object-oriented software; the technique constructs graph representations for software,
and uses these graphs to select test cases. Reference[15] presented a methodology and a tool to
support test selection from regression test suites based on change analysis in object-oriented
designs. Reference[16] proposed a technique to select only a fraction of the test cases from the
entire test suite to revalidate an object-oriented software system. Some researchers have proposed
model based regression approaches. Reference [17] proposed a test case generation technique for
UML designs using the symbolic execution method. Reference [18] presented a safe regression
technique that used various UML diagrams. Chen et al. [19] proposed a regression test selection
technique using an activity diagram of UML. Reference [20] presented a method to select the
Test Dependency Graph subset that contains a modified portion. Reference[21] shows that the
size of the specification based test-suites can be dramatically reduced, and that the fault detection
of the reduced test suites is adversely affected.
Beydeda etal.[22] suggested a technique for class level regression testing based on specification
and implementation information;regression test cases were selected by comparing two different
versions of a model described by a class specification implementation graph (CSIG).Reference
[23]reduced test cases for the EFSM (Extended Finite StateMachines) model based on
dependence analysis, and identified the difference between the original and modified model based
on the elementary modifications.
3.THE PROPOSED TECHNIQUE
This section presents a minimum test suite extraction technique for revised model in SL/SF model
based test case generation. Let Mo be the model of a specification So, and Mm be the model for the
specification Sm modified from So. In coverage based test case generation, the test suites TCo and
TCm are generated from Mo and Mm, respectively, and the test suites are partially or fully
adequate with respect to a coverage criterion, crt. The test coverage is defined as N1/N2, where
N1 is the number of test items the test suite can cover, and N2is the number of test items in a
model. A test item is a specific condition to be evaluated as TRUE or FALSE to satisfy crt. For
example, when the test criterion, crt is transition coverage, all transitions in a SL/SF model
become the test items.
Sm consists of the modified portion (MP), affected portion (AP), and unchanged portion (UP),
compared with So. MP is the portion changed directly by the modification. MP may be introduced
by modification, deletion and/or addition of specification. AP is the portion that is not changed
directly, but affected by MP. AP may or may not appear, depending on MP. UP is neither AP nor
MP. Mm is modified from Mo to adopt the modified specification.

44
For the system with the modified specification, the minimum test suite TCmin includes the test
cases to cover only MP and AP. But those for UP should not be included in TCmin, since the test
items for UP have already been tested during testing Mo with TCo. With TCmin, the test coverage
of Mm is preserved. Additionally the items deleted from Mo and their affected test items should be
covered. Usually it is not hard to find the test cases for CPin coverage based test case generation.
But it is not an easy job to extract the test cases for AP. The proposed technique aims to find
TCmin for the SL/SF model.
In SL/SF, when states and transitions are added, deleted and/or modified, Mo and Mm behave
differently. It is possible to extract the test items adequate to a certain criterion crt in the SL/SF
model, where the test items are constructed with states or/and transitions. Let the test item set for
Mi be ITMi. Figure 1 shows the basic idea of our technique, which follows. Mmand Mo are
simulated with both TCo and TCm. When simulating the models, there are several cases we need
to consider.
1) If a test item itmi appears in Mo but does not in Mm, itmi is one of the test items that has been
deleted during modelling Mm to adopt the modification. The test case ti for itmi is needed to
check whether Mmhas in fact accurately deleteditmi from Mo. Thusif ti is in TCo,ti is included in
TCmin. If ti is not in TCo, there is no way to include ti in TCmin. This is the case that TCois not
fully adequate to crt,and tihas not been generated with the test case generation method.
2) A test item itmi may appear in both Mo and Mm. If ti in TCocovers itmi in Mo, but does not
cover the test item in Mm, itemi is an item either modified or affected by the modification. Thus
ti is included in TCmin.
3) If a test item itmj appears in Mm, but does not in Mo, itmj is one of the test items added during
modeling Mm. The test case tj for itmj is needed to check whether Mm accurately has addeditmj
to Mo;and if tj is in TCm,tj is included in TCmin.
4) itmj may appear in both Mo and Mm. In this case, if tj in TCmcovers itmj in Mm, but does not
cover the test item in Mo, itemj is an item that is either modified or affected by the addition. If tj
is not in TCmin,tj is included in TCmin.
find_minimum_test_suite (Mo,Mm,ITMo,ITMm,TCo,TCm)
TCmin=empty;
for each itmi in ITMo
if (itemi is not in ITMm)
TCmin ti
else if ((itemi is in ITMm) and
((tidoes not coveritemi in ITMm)and (ti covers itemi in ITMo))
TCmin ti
for each itmj in ITMm
if (itemj is not in ITMo),
TCmin tj
else if (itemj is in ITMo) and
((tjdoes not coveritemjin ITMo) and (tj covers itemj in ITMm))
if (tj is not in TCmin)
TCmin tj
returnTCmin
Figure1. The proposed technique in pseudo code
Figures 2 and 3 show the SL/SF models used to explain the proposed technique in detail. Figure 2
shows the original model, and Figure3 its modified version. The models have two inputs, binary
A and B. State “Active_3” and two transitions R8 and R9 between “Active_3” and “Active_1” are

45
deleted from Mo, state “Active_4” is added with two new transitions R10 and R11 between
“Active_4” and “UnActive”,and R5 is modified.
Figure2. The original model Mo.
Figure3. The modified model Mm.
If the test cases are generated with respect to transition coverage, all transitions in the models
become the test items.A test case, ti is an input sequence to evaluate the transition, Ri as TRUE.
With ti,the model reaches the destination state of Ri from the initial state, OFF. We assume each
test case starts from the initial state, as in many actual applications. The inputs are initially set to
‘0’. That is, (A, B)=(0,0). Tables 1 and 2 show possible TCo and TCm, respectively, that are fully
adequate to transition coverage.
In real test case generation, the test cases are optimized after or during test case generation. By
‘optimization’ we mean that if a test case tp in a test suite is a subset of another test case tq in the
same test suite, tp is excluded from the suite. After optimization, the test cases for R1, R3, R5, R6,

46
R8need to be excluded, and four test cases should remain in the test suite. But in this example and
the next empirical study, we do not optimize test suites, since the effect of test minimization after
the optimization is not clearly shown. Thus we assume that one test case is generated for each test
item.
Table1. Test cases in TCo.
Test Item
(Transition)
Test case
R1 {(0,0),(1,0)}
R2 {(0,0),(1,0),(0,0)}
R3 {(0,0),(1,0),(1,1)}
R4 {(0,0),(1,0),(1,1),(1,0)}
R5 {(0,0),(1,0),(1,1),(1,1)}
R6 {(0,0),(1,0),(1,1),(1,1),(1,2)}
R7 {(0,0),(1,0),(1,1),(1,1),(1,2),(1,0}
R8 {(0,0),(1,0),(1,1),(1,3)}
R9 {(0,0),(1,0),(1,1),(1,3),(1,0)}
Table2. Test cases in TCm.
Test Item
(Transition)
Test case
R1 {(0,0),(1,0)}
R2 {(0,0),(1,0),(0,0)}
R3 {(0,0),(1,0),(1,1)}
R4 {(0,0),(1,0),(1,1),(1,0)}
R5 {(0,0),(1,0),(1,1),(1,1)}
R6 {(0,0),(1,0),(1,1),(1,1),(1,1),(1,2)}
R7 {(0,0),(1,0),(1,1),(1,1),(1,1),(1,2),(1,0}
R10 {(0,0),(1,0),(1,2)}
R11 {(0,0),(1,0),(1,2),(1,0)}
First, the test items in ITMo are compared with those in ITMm. R8 and R9 in ITMo are not in ITMm.
Thus t8,o and t9,o, the test cases for R8 and R9 in Mo, are inserted in TCmin. tm,n is defined as the nth
test case in model Mn. The two cases are needed to check the correctness of deletion. R1 through
R7 appear in both Mo and Mm. By simulating the models with TCo,R1 through R5are found
covered by t1,o through t5,oin both models. But R6 and R7 in Mm are not evaluated as TRUE with
t6,o and t7,o; mean while, those in Moare covered. This means that there is one or more
modifications and/or affected portion sin R6 and R7 by deleting R8 and R9. In this case, there is an
effect. Thus t6,o through t7,o are put in TCmin. Table 3 shows the simulation result of Mm with TCo.
The simulation result of Mo is not summarized, since all items are covered.
Now compare the test items in ITMm, with those in ITMo. R10 and R11 in ITMm do not appear in
ITMo. Thus t10,m and t11,m, which are needed to check the specification addition, are inserted into
TCmin. The result of simulating the models with TCm reveals that all test items in both models are
satisfied. Thus, no extra test cases are inserted intoTCmin. Finally,TCmin for the model in Figure3
is determined as {t8,o, t9,o, t6,o, t7,o, t10,m, t11,m}.Table4 shows the result of simulating Mo with TCm.
The simulation result of Mm is not included.

47
Table3. Result of simulating Mm with TCo.
Test Items TCo Criterion
satisfactionITMo ITMm Test cases ti,j
R1 R1 {(0,0),(1,0)} t1,o O
R2 R2 {(0,0),(1,0),(0,0)} t2,o O
R3 R3 {(0,0),(1,0),(1,1)} t3,o O
R4 R4 {(0,0),(1,0),(1,1),(1,0)} t4,o O
R5 R5 {(0,0),(1,0),(1,1),(1,1)} t5,o O
R6 R6 {(0,0),(1,0),(1,1),(1,1),(1,2)} t6,o X
R7 R7 {(0,0),(1,0),(1,1),(1,1),(1,2),(1,0} t7,o X
R8 - {(0,0),(1,0),(1,1),(1,3)} t8,o NA
R9 - {(0,0),(1,0),(1,1),(1,3),(1,0)} t9,o NA
Table4. Result of simulating Mo with TCm.
Test Items TCm Criterion
satisfactionITMo ITMm Test cases ti,j
R1 R1 {(0,0),(1,0)} t1,m O
R2 R2 {(0,0),(1,0),(0,0)} t2,m O
R3 R3 {(0,0),(1,0),(1,1)} t3,m O
R4 R4 {(0,0),(1,0),(1,1),(1,0)} t4,m O
R5 R5 {(0,0),(1,0),(1,1),(1,1)} t5,m O
R6 R6 {(0,0),(1,0),(1,1),(1,1),(1,1),(1,2)} t6,m O
R7 R7 {(0,0),(1,0),(1,1),(1,1),(1,1),(1,2),(1,0} t7,m O
- R10 {(0,0),(1,0),(1,2)} t10,m NA
- R11 {(0,0),(1,0),(1,2),(1,0)} t11,m NA
4.EMPIRICAL STUDY
The proposed technique is verified using two ECU (Electrical Control Unit) SL/SF models for a
commercial vehicle provided by a world class automobile manufacturer. One ECU is the
Intelligent Management System (IMS), which is used for managing the ECUs for doors, mirrors,
and wipers. The other is the driver side mirror ECU (DDM). The IMS model is simple, and has
34 states and 51 transitions; meanwhile, the mirror model is relatively complicated, and has 49
states and 135 transitions.
For the study, we use the models modified according to the three different modification types:
transition modification, state/transition addition, and state/transition deletion. The modifications
were not made arbitrarily, but were made while keeping the precise specifications of the car.
First, the test suites for the original IMS (TCOI), modified IMS (TCMI), original DDM (TCOD), and
modified DDM (TCMD) models were generated with a commercial test case generator, TCG [24].
The generated test suites were partially adequate to transition coverage, due to various reasons,
such as incomplete model, poor performance of TCG, and unknown reasons. We performed an
empirical study with the test suites. Since the test case generation is beyond the scope of this
paper, we do not discuss the test case generation from the model. Using the original and modified
models, and their test suites, the proposed technique tries to find the minimized test suites. The
test cases in TCmin are compared with those of minimum test cases analyzed by the inspection
with respect to the number of test cases for the modification itself (Direct in Table5), and that for
the affected portion (Affected in Table5).

48
Table5 is a summary of the study. In the IMS model, one transition is modified among 51
transitions in the Modification type. Our technique finds TCmin has just one test case for Direct
test cases, but no Affected test cases. This indicates that there is no portion affected by the
modification. We confirm it by inspecting the model. One state and the three related transitions
are added in the Addition type. It is found that TCmin includes three Direct test cases needed to
check the modified transitions. As in the Modification type, no extra Affected test cases are
needed. In the Deletion type experiment, one state and the connected three transitions are deleted
from IMSo. Like the previous two experiments, the test suite includes only three Direct test cases
to check the deleted transitions. For IMS, TCmin consists of only Direct test cases in all three
modification types. We confirmed that the test suites were minimum, and sufficient to cover the
model change. Since the IMS model is small, and consists of seven almost independent sub-
models that don’t affect each other, the modification does not affect other parts, and TCmin
contains the test cases to cover only the modified transitions.
The DDM model is relatively complicated, and some parts are closely related to each other. In the
Modification type experiment, two out of 135 transitions are modified. By the modifications, 36
transitions are affected. The proposed technique finds two Direct and 36 Affected test cases for
TCmin. Three transitions are added, and the added transitions affect three other transitions in the
Addition type. In this case, the technique finds the minimized test suite that includes six test
cases. In the Deletion type, three transitions are deleted with one state. In this case, TCmin consists
of three Direct and three Affected test cases. TCmin is minimum in all three modification types.
Table5. Summary of the empirical study.
Model
Modification
Type
No. of
modified
test items
No. of minimum test case
(No. of found test cases)
Direct Affected
IMS
Modification 1 1(1) 0
Addition 3 3(3) 0
Deletion 3 3(3) 0
DDM
Modification 2 2(2) 36(36)
Addition 3 3(3) 3(3)
Deletion 3 3(3) 3(3)
The empirical study is limited in terms of the number of models and modification complexity.
But we believe that it demonstrates the effectiveness of the proposed technique, even though it is
not sufficient to fully verify the performance.
Sometimes test engineers are forced to test a modified system with test cases only for the
modified portion because of limited test time, losing the coverage. Without the reduced test suite,
it is inevitable that the modified model has to be tested with the test suite with all test cases
generated by the test case generator to preserve the original coverage. Even with the test cases, it
is not possible to test the deleted portion that does not appear in the modified model. Table6
summarizes the statistics of test cases generated by TCG, and the minimized test cases found by
the proposed technique. Even though the test case reduction surely varies with model and
modification, the statistics demonstrate that the reduction in the number of test cases may be
significant in some applications, while preserving the coverage of the full test suite.

49
Table6.Test case statistics.
5.CONCLUSIONS
We have described a minimized test case extraction technique for a modified SL/SF model. The
proposed technique is capable of extracting a minimized sized test suite by taking into account
both the modified and the affected portion of the revised model. For the empirical study, we used
two models for the ECUs deployed in a commercial vehicle. The result of our empirical study
shows that the proposed technique achieved a significant reduction in terms of test cases. It
extracts just the test cases needed for testing the portion modified from the original model, and
those affected by the modification. The cost of testing was dramatically reduced for the models
used in this study, even though the performance and cost reduction are surely dependent on the
models and modifications. Although we cannot broadly generalize our results, and further studies
are needed, the experiment indicates that the proposed technique may be an effective means of
reducing testing effort.
ACKNOWLEDGEMENTS
This Research was supported by Defense Acquisition Program Administration(UD150042AD).
REFERENCES
[1] H. Leung and L. White, (1990) “A Study of Integration Testing and Software Regression at the
Integration Level”, Proceedings of Conference on Software
Maintenance,DOI:10.1109/ICSM.1990.131377, pp290-301.
[2] D. Nardo, N. Alshahwa1, L. Briand, (2013) “Coverage-Based Test Case Prioritisation: An Industrial
Case Study”, Proceedings of IEEE Sixth International Conference on Software Testing.
[3] H. Agrawal, J.R. Horgan, E.W. Krauser and S. London, (1993) “Incremental Regression Testing”,
Proceedings of IEEE Conference on Software Maintenance, DOI: 10.1109/ICSM.1993.366927,
pp348-357.
[4] D. Deng, P.C.Y. Sheu and T. Wang, (2004) “Model-based Testing and Maintenance”,Proceedings of
IEEE Sixth International Symposium on Multimedia Software Engineering, DOI:
10.1109/MMSE.2004.51,pp278-285.
[5] C.S. Pasareanu, J. Schumann, P. Mehlitz, M. Lowry, G. Karsai, H. Nine and S. Neema,(2009) "Model
Based Analysis and Test Generation for Flight Software", Proceedings of IEEE Third International
Conference onSpace Mission Challenges for Information Technology, DOI: 10.1109/SMC-IT.2009.18,
pp83-90.
[6] Lutz Köster, Thomas Thomsen, and Ralf Stracke, (2001) “Connecting Simulink to OSEK: Automatic
Code Generation for Real-Time Operating Systems with TargetLink”, SAE Technical Paper.
[7] Reactis, <http://www.reactive-systems.com/papers/bcsf.pdf>.
[8] S. Yoo and M. Harman, (2012) “Regression Testing Minimization Selection and prioritization:
Survey”, Software TestingVerification & Reliability, Vol. 22, Issue 2, DOI: 10.1002/stvr.430, pp67-
120.
Model Modification
Type
No. of
test cases
(TCG)
No. of
minimized test cases
(the proposed technique)
IMS
Modification 49 1
Addition 52 3
Deletion 46 3
DDM
Modification 102 38
Addition 105 6
Deletion 99 6

50
[9] A. Leitner, M. Oriol, and A. Zeller, (2007) “Efficient Test Case Minimization”, Proceedings of the
twenty-second IEEE/ACM international conference on Automated software engineering, DOI:
10.1145/1321631.1321698, pp417-420.
[10] J. Lin, C. Huang and C. Lin, (2008) “Test suite reduction analysis with enhanced tie-breaking
techniques”, Management of Innovation and Technology, Proceedings of IEEE4th International
Conference on, DOI: 10.1109/ICMIT.2008.4654545, pp1228-1233.
[11] S. Elbaum, A. Malishevsky, and G. Rothermel, (2001) “Incorporating Varying Test Costs and Fault
Severities into Test Case Prioritization”,Proceedings of the 23rd International Conference on Software
Engineering,DOI: 10.1109/ICSE.2001.919106, pp329-338.
[12] Z. Li, M. Harman and R. Hierons, (2007) “Search Algorithms for Regression Test Case Prioritization”,
Proceedings of the IEEE Transactions on Software Engineering, Vol. 33, Issue 4, DOI:
10.1109/TSE.2007.38, pp225-237.
[13] G. Rothermel and M. Harrold, (1994) “Selecting tests and identifying test coverage requirements for
modified software”, Proceedings of the 1994 ACM SIGSOFT international symposium on Software
testing and analysis,DOI: 10.1145/186258.187171, pp169-184.
[14] G. Rothermel, M. Harrold and J. Dedhia, (2000) “Regression test selection for C++ software”,
Software Testing Verification and Reliability, Volume 10, Issue 2, DOI: 10.1002/1099-
1689(200006)10:2<77::AID-STVR197>3.0.CO;2-E, pp77-109.
[15] L.C. Briand, Y. Labiche and S. He, (2009) “Automating regression test selection based on UML
designs”, Information and Software Technology, Volume 51, Issue 1, DOI:
10.1016/j.infsof.2008.09.010, pp16-30.
[16] P. Hsia, X. Li, D. Kung, C. Hsu, L. Li, Y. Toyoshima and C. Chen, (1997) “A technique for the
selective revalidation of object-oriented software”, Journal of Software Maintenance: Research and
Practice, Volume. 9, Issue 4, DOI: 10.1002/(SICI)1096-908X(199707/08)9:4<217::AID-
SMR152>3.0.CO;2-2., pp217-233.
[17] L. Briand, Y. Labiche, and T. Yue, (2009) “Automated Traceability Analysis for UML Model
Refinements”, Information and Software Technology, Volume 51, Issue 2, DOI:
10.1016/j.infsof.2008.06.002, pp. 512-527.
[18] L. Briand,Y. Labiche, and G. Soccar, (2002)“Automating Impact Analysis and Regression Test
Selection Based on UML Designs”,Proceedings. International Conference on Software Maintenance,
DOI: 10.1109/ICSM.2002.1167775, pp. 252-261.
[19] Y. Chen, R.L. Probert, and D.P. Sims,(2002) “Specification-based Regression Test Selection with Risk
Analysis”, Proceedings of the 2002 conference of the Centre for Advanced Studies on Collaborative
research, DOI: 10.1145/782115.782116, pp1.
[20] Y. Traon, T. Jeron, J.M. Jezequel, and P. Morel, (2000) “Efficient object-oriented integration and
regression testing”, Proceedings of the IEEE Transactions on Reliability, Vol. 49, Issue 1, DOI:
10.1109/24.855533, pp12-25.
[21] M. Heimdahl and D. George, (2004) “Test-Suite Reduction for Model Based Tests: Effects on Test
Quality and Implications for Testing”, Proceedings of the 19th IEEE international conference on
Automated software engineering, DOI: 10.1109/ASE.2004.67, pp176-185.
[22] S. Beydeda and V. Gruhn, (2001) “Integrating White- and Black-Box Techniques for Class-Level
Regression Testing”, Computer Software and Applications Conference, COMPSAC 2001. 25th
Annual International, DOI: 10.1109/CMPSAC.2001.960639, pp357–362.
[23] B. Korel, L.H. Tahat, and B. Vaysburg, (2002) “Model Based Regression Test Reduction Using
Dependence Analysis”, Proceedings. International Conference onSoftware Maintenance, DOI:
10.1109/ICSM.2002.1167768, pp214-223.
[24] Btstech, <http://www.btstech.co.kr/page_vaHk97>.

51
AUTHORS
Han Gon Park holds a Master Degree (M.S) in Electronic Engineering from Ajou University,
Republic of Korea. His areas of research interest includes Embedded System Testing, Model
Based Testing. He is currently working at LG Chem.
Geon Gu Park is a Master’s course of Electronic Engineering from Ajou University, Republic
of Korea. His areas of research interest includes Embedded System Testing, Model Based
Testing.
Kihyun Chung holds a Doctoral Degree (Ph.D) in Electronic Engineering from Purdue
University, USA. His areas of research interest includes Computer Architecture, VLSI, Real
time/ Multimedia System. At present he is working as Professor, Electronic Engineering, Ajou
University, Republic of Korea.
Kyunghee Choi holds a Doctoral Degree (Ph.D) in Computer Engineering from Paul Sabatier
University, France. His areas of research interest includes Operating System, Real time/
Multimedia System. At present he is working as Professor, Computer Engineering, Ajou
University, Republic of Korea.

EXTRACTING THE MINIMIZED TEST SUITE FOR REVISED SIMULINK/STATEFLOW MODEL

More Related Content

What's hot (19)

Similar to EXTRACTING THE MINIMIZED TEST SUITE FOR REVISED SIMULINK/STATEFLOW MODEL (20)

Recently uploaded (20)

EXTRACTING THE MINIMIZED TEST SUITE FOR REVISED SIMULINK/STATEFLOW MODEL