Regression Verification in a User-Centered Software
Development Process for Evolving Automated Production
The vision for this project is to advance technology such that
regression verification methods are available that will be
routinely used for ensuring correctness in the evolution
processes for long-living reliable systems requiring frequent
re-validation. The goal of regression verification is to
formally prove that software remains correct through its
evolution, changes have the desired effect, and no new bugs
Regression verification avoids the main
bottleneck for the routine practical use of formal
verification, namely the need to write full functional
specifications (which is a huge effort). Also, given two
program versions or variants that are both complex but similar
to each other, the effort for verifying the relation between
them mainly depends on the difference between the programs and
not on their overall size and complexity.
As a follow-up to the IMPROVE project, we plan to overcome the two remaining obstacles on the path to reaching the above vision of routine use:
- Reach and power: The reach and power of regression verification needs to be extended to cover real-world systems and change scenarios.
- User in the loop: Regression verification needs to be integrated into the software development and evolution process, and useful feedback needs to be given to the user in case a verification attempt is not immediately successful.
In this phase, we will target a particular application domain. This provides the third main motif for our project in Phase 2 – now called IMPROVE APS:
Automated production systems in the pharmaceutical and the food-manufacturing industry: The control software and discrete processes of automated production systems (aPS) are a very promising application area for regression verification.
These systems are long-living and have to fulfill dependability and reliability criteria to avoid machine hazards or accidents involving operator or maintenance personnel.
Applicability in realistic industrial situations is within reach as we focus on evolution with limited changes (e.g., bug fixes) and small variations in the software.
As case studies, we use manufacturing systems in the pharmaceutical and the food industry, where aPS must be validated following regulations to ensure no harm is inflicted on consumers of the products.
The envisaged solutions of IMPROVE APS follow the goal of a comprehensive elaboration of approaches to manage software evolution with additional constraints from application taken into account.
This project is part of the DFG SPP1593 "Design for Future"
This project is the successor of
which focused on regression verification of imperative and object-oriented programs.
We have investigated the foundations of regression verification, developed the basic methods, and shown their applicability to realistic examples and case studies.
In line with the new focus on integration into the software development process and the application domain of automated production systems, IMPROVE APS has a new set of principal investigators and is now a collaboration of the Institute of Automation and Information Systems
(led by Birgit Vogel-Heuser
) at Technische Universität München (TUM) and the Application-oriented Formal Verification
group (led by Bernhard Beckert
) at Karlsruhe Institute of Technology (KIT).
|Generalised Test Tables: A Practical Specification Language for Reactive Systems||Bernhard Beckert, Suhyun Cha, Mattias Ulbrich, Birgit Vogel‑Heuser, and Alexander Weigl||13th International Conference on integrated Formal Methods (iFM 2017)|
|Generation of Monitoring Functions in Production
Automation Using Test Specifications||Suhyun Cha, Sebastian Ulewicz, Birgit Vogel‑Heuser, Alexander Weigl, Mattias Ulbrich, and Bernhard Beckert||15th IEEE International Conference on
Industrial Informatics (INDIN 2017)|
|Generalized Test Tables: A Powerful and Intuitive
Specification Language for Reactive Systems||Alexander Weigl, Franziska Wiebe, Mattias Ulbrich, Sebastian Ulewicz, Suhyun Cha, Michael Kirsten, Bernhard Beckert, and Birgit Vogel‑Heuser||15th IEEE International Conference on Industrial
Informatics (INDIN 2017)|
|Regression Verification for Programmable Logic Controller Software||Bernhard Beckert, Mattias Ulbrich, Birgit Vogel‑Heuser, and Alexander Weigl||17th International Conference on Formal Engineering Methods (ICFEM 2015)|
|Proving Equivalence between Control Software Variants for Programmable Logic Controllers -
Using Regression Verification to Reduce Unneeded Variant Diversity||Sebastian Ulewicz, Mattias Ulbrich, Alexander Weigl, Bernhard Beckert, and Birgit Vogel‑Heuser||20th IEEE International Conference on Emerging Technologies and Factory Automation (ETFA 2015)|
|Selected Challenges of Software Evolution for Automated Production Systems||Birgit Vogel‑Heuser, Stefan Feldmann, Jens Folmer, Matthias Kowal, Ina Schaefer, Jan Ladiges, Alexander Fay, Christopher Haubeck, Winfried Lamersdorf, Sascha Lity, Timo Kehrer, Matthias Tichy, Sinem Getir, Mattias Ulbrich, Vladimir Klebanov, and Bernhard Beckert||13th IEEE International Conference on Industrial Informatics (INDIN 2015)|
|Regression Verification for Java Using a Secure Information Flow Calculus||Bernhard Beckert, Vladimir Klebanov, and Mattias Ulbrich||17th Workshop on Formal Techniques for
Java-like Programs (FTfJP 2015)|
|Regression Verification for Programmable Logic Controller Software||Bernhard Beckert, Mattias Ulbrich, Birgit Vogel‑Heuser, and Alexander Weigl||Karlsruhe Institute of Technology, Department of Informatics 2015-06|
|Automating Regression Verification||Dennis Felsing, Sarah Grebing, Vladimir Klebanov, Philipp Rümmer, and Mattias Ulbrich||Multikonferenz Software Engineering und Management 2015:
Fachtagung Software Engineering (SE 2015)|
|Regression Verification for Programmable Logic
Controller Software||Alexander Sebastian Weigl||Karlsruhe Institute of Technology (January 2015)|
On request, we provide the toolchain presented at ICFEM '15
Feel free to contact Mattias Ulbrich
For more information, please have a look at ReVe for PLC
Additionally, most software parts associated to our publications are open source; see VerifAPS