The document discusses executable modeling and simulation of system software and processes. It describes how modeling languages like statecharts, Petri nets, and BPMN can be used to model control software, business processes, and other complex systems. Models can then be simulated and analyzed to better understand problems, reduce complexity, detect errors early, and compare alternative solutions. Challenges discussed include providing better support for domain-specific modeling, model-based testing, and executable robot modeling.

1. Tom Mens
tom.mens@umons.ac.be
Executable modeling and simula2on
of system so5ware and processes
So1ware Engineering Lab, Dept. Computer Science

2. Université de Mons
So5ware-controlled systems
are omnipresent
Prof. Tom Mens Service de Génie Logiciel Département d’InformaEque

3. Université de Mons
Control so1ware can be very complex
§ ConEnuous interacEon between so1ware and hardware
§ ConEnuous interacEon with external world and users
§ Must respect func%onal requirements
§ Vending machine does not dispense correct beverage
§ Cash machine returns more cash than requested
§ …
§ Must respect non-func%onal requirements
§ Safety / Security / Performance / Energy constraints / Maintainability /
Usability / …
§ Microwave/elevator should not funcEon with open doors
§ Traffic lights should never be green simultaneously
So5ware-controlled systems
are difficult to develop
3 Prof. Tom Mens Service de Génie Logiciel Département d’InformaEque

4. Université de Mons
Example 1:
Modèle du processus
d’engagement de personnel
UMONS (version 1, 2013)
The same is true for
(business) processes
4 Prof. Tom Mens Service de Génie Logiciel Département d’InformaEque

5. Université de Mons
Example 2:
Distributed software
development and bug
fixing processes
The same is true for
(business) processes
5 Prof. Tom Mens Service de Génie Logiciel Département d’InformaEque

6. Université de Mons
• beUer understand the problem
• reduce the “accidental complexity” of the
soluEon
• detect errors early
• Test or verify the soluEon before it is actually implemented
• explore the design space
• compare alternaEve soluEons through simulaEon
Model simula2on and analysis
can help to …
6 Prof. Tom Mens Service de Génie Logiciel Département d’InformaEque

7. Université de Mons
• Allow to express the soluEon at a higher level of
abstracEon than tradiEonal programming
languages
• Provide support for
• Simula%ng the desired behaviour or process
• Genera%ng so1ware code automaEcally from the simulated
model in order to
• Execute and integrate the generated so1ware with other so1ware or
hardware
• Support or control the modeled process
• Verifying correctness of the modeled behaviour
Modeling languages
7 Prof. Tom Mens Service de Génie Logiciel Département d’InformaEque

8. Université de Mons
Modeling Languages
Example : Statecharts
8 Prof. Tom Mens Service de Génie Logiciel Département d’InformaEque

9. Université de Mons
Many tools for simulaEon and execuEon
Example: Yakindu (www.statecharts.org)
Modeling Tools
Example : Statecharts
9 Prof. Tom Mens Service de Génie Logiciel Département d’InformaEque

10. Université de Mons
SISMIC (created by A. Decan)
A tool under development at So1ware Engineering Lab (UMONS) to
• Simulate statecharts
• Integrate with Python code
• Facilitate tesEng
(based on “stories”)
• Support mulEple
communicaEng statecharts
• Express statechart contracts
(invariants, pre- and postcondiEons)
• And many more…
See hUps://github.com/AlexandreDecan/sismic
Modeling Languages
Example : Statecharts
10 Prof. Tom Mens Service de Génie Logiciel Département d’InformaEque

11. Université de Mons
Modeling Languages
Example : Petri nets
11 Prof. Tom Mens Service de Génie Logiciel Département d’InformaEque
See “Petri Nets World” for more information
http://www.informatik.uni-hamburg.de/TGI/PetriNets

12. Université de Mons
Many tools for simulaEon and formal analysis
Example: Pipe2 (pipe2.sourceforge.net)
Modeling tools
Example: Petri nets
12 Prof. Tom Mens Service de Génie Logiciel Département d’InformaEque

13. Université de Mons
Modeling languages
Examples: BPMN
13 Prof. Tom Mens Service de Génie Logiciel Département d’InformaEque

14. Université de Mons
Example:
Visual Paradigm
Simulate execuEon of business process to
• Study resource consumpEon (e.g. human resources, devices)
throughout the process
• Evaluate cost
• IdenEfy boUlenecks
• Compare design alternaEves
Modeling tools
BPMN
14 Prof. Tom Mens Service de Génie Logiciel Département d’InformaEque

15. Université de Mons
Model verifica2on and model checking
Verify if the model has all
desirable proper2es
(structural, behavioural,
temporal, …)
Use most appropriate
formalism
Use most appropriate
(model checking) tool
Examples
• Reachability
• Safety: something should never
happen
• Liveness: something must eventually
happen
• Fairness: every possible process
should be executed infinitely o1en
• linear temporal logic (LTL)
• computaEon tree logic (CTL)
• …
• SPIN
• Alloy
• …
15 Prof. Tom Mens Service de Génie Logiciel Département d’InformaEque

16. Université de Mons
Provide beUer tool support and formal support for
• Domain-specific modeling
• Express models in a language close to the domain expert
• Human-computer interacEon modeling
• E.g. to control robots
• Model-based tesEng
• Facilitate tesEng of models / generate automated tests from models
• Design space exploraEon
• Evaluate the qualiEes of alternaEve models
• Model evoluEon
• Facilitate changing the model while preserving its desirable properEes
Challenges
16 Prof. Tom Mens Service de Génie Logiciel Département d’InformaEque

17. Université de Mons
Example: A modeling language for gestural interacEon
(GISMO, PhD thesis Romuald Deshayes, UMONS, 2015)
Challenges
Domain-specific so5ware modeling
17 Prof. Tom Mens Service de Génie Logiciel Département d’InformaEque

18. Université de Mons
Example: A modeling language for gestural interacEon
(GISMO, PhD thesis Romuald Deshayes, UMONS, 2015)
Challenges
Domain-specific so5ware modeling
18 Prof. Tom Mens Service de Génie Logiciel Département d’InformaEque

19. Université de Mons
Challenges
Executable robot modeling
19
Mémoire
Christophe D’Hondt

20. Université de Mons
• Research in progress
• Property-based tesEng
• Generate simplest test cases that violate a given property / invariant
• Design by contract
• Express pre- and postcondiEons and invariants on the model
• Raise excepEons during simulaEon/execuEon if contract violated
• Test generaEon
• Use model to generate tests for source code automaEcally
• MutaEon tesEng
• To evaluate and improve exisEng test suite
Challenges
Model-based tes2ng
20