Towards Modular Verification of Software Product Lines with mCRL2

Lecture Notes in Computer Science

Reniers

Vink

2016

Self Cite

Abstract. Using the CIF 3 toolset, we illustrate the general idea of controller synthesis for product line engineering for a prototypical example of a family of coffee machines. The challenge is to integrate a number of given components into a family of products such that the resulting behaviour is guaranteed to respect an attributed feature model as well as additional behavioural requirements. The proposed correctness-byconstruction approach incrementally restricts the composed behaviour by subsequently incorporating feature constraints, attribute constraints and temporal constraints. The procedure as presented focusses on synthesis, but leaves ample opportunity to handle e.g. uncontrollable behaviour, dynamic reconfiguration, and product-and family-based analysis.

“…1 yields a product automaton of all features together with a supervisor which has 16 initial states, each corresponding to a valid product. This is the same number as reported in [4,5].…”

Section: Requirement Invariant Fb Present ⇔ ⇔ ⇔ Fc Present ;supporting

confidence: 87%

Section: Component Behaviourmentioning

confidence: 99%

See 1 more Smart Citation

Supervisory Controller Synthesis for Product Lines Using CIF 3

Lecture Notes in Computer Science

Reniers

Vink

2016

Self Cite

“…org) for inspiration, since it allows to model actions parametrized with userdefined abstract datatypes and to verify formulas in the modal μ-calculus, thus allowing to quantify over data [25]. Moreover, also mCRL2 is recently being used for product family analysis [10][11][12].…”

Section: Discussionmentioning

confidence: 99%

Model Checking Value-Passing Modal Specifications

Lecture Notes in Computer Science

Gnesi

Mazzanti

2015

Self Cite

Abstract. Formal modelling and verification of variability concepts in product families has been the subject of extensive study in the literature on Software Product Lines. In recent years, we have laid the basis for the use of modal specifications and branching-time temporal logics for the specification and analysis of behavioural variability in product family definitions. A critical point in this formalization is the lack of a possibility to model an adequate representation of the data that may need to be described when considering real systems. To this aim, we now extend the modelling and verification environment that we have developed for specifications interpreted over Modal Transition Systems, by adding the possibility to include data in the specifications. In concert with this, we also extend the variability-specific modal logic and the associated special-purpose model checker VMC. As a result, it offers the possibility to efficiently verify formulas over possibly infinite-state systems by using the on-the-fly bounded model-checking algorithms implemented in the model checker. We illustrate our approach by means of a simple yet intuitive example: a bike-sharing system.

“…In [7,8,10], we showed how the formal specification language mCRL2 and its industrial-strength toolset can be exploited to model and analyze SPL. The mCRL2 toolset supports parametrized modeling, model reduction and quality assurance techniques like model checking.…”

Section: Introductionmentioning

confidence: 99%

Towards a Feature mu-Calculus Targeting SPL Verification

Electron. Proc. Theor. Comput. Sci.

Vink

Willemse

2016

Self Cite

The modal µ-calculus µL is a well-known fixpoint logic to express and model check properties interpreted over labeled transition systems. In this paper, we propose two variants of the µ-calculus, µL f and µL ′ f , for feature transition systems. For this, we explicitly incorporate feature expressions into the logics, allowing operators to select transitions and behavior restricted to specific products and subfamilies. We provide semantics for µL f and µL ′ f and relate the two new µ-calculi and µL to each other. Next, we focus on the analysis of SPL behavior and show how our formalism can be applied for product-based verification with µL f as well as family-based verification with µL ′ f . We illustrate by means of a toy example how properties can be model checked, exploiting an embedding of µL ′ f into the µ-calculus with data.