Johannes Bürdek

Reasoning about product-line evolution using complex feature model differences

Features define common and variable parts of the members of a (software) product line. Feature models are used to specify the set of all valid feature combinations. Feature models not only enjoy an intuitive tree-like graphical syntax, but also a precise formal semantics, which can be denoted as propositional formulae over Boolean feature variables. A product line usually constitutes a long-term investment and, therefore, has to undergo continuous evolution to meet ever-changing requirements. First of all, product-line evolution leads to changes of the feature model due to its central role in the product-line paradigm. As a result, product-line engineers are often faced with the problems that (1) feature models are changed in an ad-hoc manner without proper documentation, and (2) the semantic impact of feature diagram changes is unclear. In this article, we propose a comprehensive approach to tackle both challenges. For (1), our approach compares the old and new version of the diagram representation of a feature model and specifies the changes using complex edit operations on feature diagrams. In this way, feature model changes are automatically detected and formally documented. For (2), we propose an approach for reasoning about the semantic impact of diagram changes. We present a set of edit operations on feature diagrams, where complex operations are primarily derived from evolution scenarios observed in a real-world case study, i.e., a product line from the automation engineering domain. We evaluated our approach to demonstrate its applicability with respect to the case study, as well as its scalability concerning experimental data sets.

Staged configuration of dynamic software product lines with complex binding time constraints

Dynamic software product lines (DSPL) constitute a promising approach for developing highly-configurable, runtime-adaptive systems in a feature-oriented way. A DSPL integrates both variability in time and space in a unified conceptual framework. For this, domain features are equipped with additional binding time information to distinguish between static configuration parameters and dynamically (re-) configurable features. Until now, little support exists to specify and validate staged (re-)configuration semantics for DSPLs in a concise way. In this paper, we propose conservative extensions to domain feature models comprising variable feature binding times together with different kinds of binding time constraints. Those extensions are motivated by a real-world industrial case study from the automation engineering domain. Our implementation performs a model transformation into plain feature models treatable by corresponding state-of-the-art analysis tools. We conducted an evaluation of our approach concerning the case study.

Applying Model-based Software Product Line Testing Approaches to the Automation Engineering Domain

Abstract The software constitutes a major part of nowadays automation systems being responsible for conducting complex control tasks. Machines and plants are often unique in some industrial branches; hence, they become mechatronic products configured individually. The inherent software variability of those highly-configurable systems makes efficient, yet accurate quality assurance a challenging task. This article presents a comprehensive approach for applying model-based software product line testing techniques to the automation engineering domain. Existing approaches for variability modeling are adapted to domain specific modeling languages to allow for variability-aware test case generation and execution. The implementation of the approach is evaluated by means of a sample automation system product line.

Fault-based product-line testing: effective sample generation based on feature-diagram mutation

Testing every member of a product line individually is often impracticable due to large number of possible product configurations. Thus, feature models are frequently used to generate samples, i.e., subsets of product configurations under test. Besides the extensively studied combinatorial interaction testing (CIT) approach for coverage-driven sample generation, only few approaches exist so far adopting mutation testing to emulate faults in feature models to be detected by a sample. In this paper, we present a mutation-based sampling framework for fault-based product-line testing. We define a comprehensive catalog of atomic mutation operators on the graphical representation of feature models. This way, we are able (1) to also define complex mutation operators emulating more subtle faults, and (2) to classify operators semantically, e.g., to avoid redundant and equivalent mutants. We further introduce similarity-based mutant selection and higher order mutation strategies to reduce testing efforts. Our implementation is based on the graph transformation engine Henshin and is evaluated concerning effectiveness/efficiency trade-offs.

Facilitating Reuse in Multi-goal Test-Suite Generation for Software Product Lines

Software testing is still the most established and scalable quality-assurance technique in practice. However, generating effective test suites remains computationally expensive, consisting of repetitive reachability analyses for multiple test goals according to a coverage criterion. This situation is even worse when testing entire software product lines, i.e., families of similar program variants, requiring a sufficient coverage of all derivable program variants. Instead of considering every product variant one-by-one, family-based approaches are variability-aware analysis techniques in that they systematically explore similarities among the different variants. Based on this principle, we present a novel approach for automated product-line test-suite generation incorporating extensive reuse of reachability information among test cases derived for different test goals and/or program variants. We present a tool implementation on top of CPA/tiger which is based on CPAchecker, and provide evaluation results obtained from various experiments, revealing a considerable increase in efficiency compared to existing techniques.

Specification and automated validation of staged reconfiguration processes for dynamic software product lines

Dynamic software product lines (DSPLs) propose elaborated design and implementation principles for engineering highly configurable runtime-adaptive systems in a sustainable and feature-oriented way. For this, DSPLs add to classical software product lines (SPL) the notions of (1) staged (pre-)configurations with dedicated binding times for each individual feature, and (2) continuous runtime reconfigurations of dynamic features throughout the entire product life cycle. Especially in the context of safety- and mission-critical systems, the design of reliable DSPLs requires capabilities for accurately specifying and validating arbitrary complex constraints among configuration parameters and/or respective reconfiguration options. Compared to classical SPL domain analysis which is usually based on Boolean constraint solving, DSPL validation, therefore, further requires capabilities for checking temporal properties of reconfiguration processes. In this article, we present a comprehensive approach for modeling and automatically verifying essential validity properties of staged reconfiguration processes with complex binding time constraints during DSPL domain engineering. The novel modeling concepts introduced are motivated by (re-)configuration constraints apparent in a real-world industrial case study from the automation engineering domain, which are not properly expressible and analyzable using state-of-the-art SPL domain modeling approaches. We present a prototypical tool implementation based on the model checker SPIN and present evaluation results obtained from our industrial case study, demonstrating the applicability of the approach.

Measuring effectiveness of sample-based product-line testing

Recent research on quality assurance (QA) of configurable software systems (e.g., software product lines) proposes different analysis strategies to cope with the inherent complexity caused by the well-known combinatorial-explosion problem. Those strategies aim at improving efficiency of QA techniques like software testing as compared to brute-force configuration-by-configuration analysis. Sampling constitutes one of the most established strategies, defining criteria for selecting a drastically reduced, yet sufficiently diverse subset of software configurations considered during QA. However, finding generally accepted measures for assessing the impact of sample-based analysis on the effectiveness of QA techniques is still an open issue. We address this problem by lifting concepts from single-software mutation testing to configurable software. Our framework incorporates a rich collection of mutation operators for product lines implemented in C to measure mutation scores of samples, including a novel family-based technique for product-line mutation detection. Our experimental results gained from applying our tool implementation to a collection of subject systems confirms the widely-accepted assumption that pairwise sampling constitutes the most reasonable efficiency/effectiveness trade-off for sample-based product-line testing.

Modeling and Testing Product Lines with Unbounded Parametric Real-Time Constraints

Real-time requirements are crucial for embedded software in many modern application domains of software product lines. Hence, techniques for modeling and analyzing time-critical software have to be lifted to software product line engineering, too. Existing approaches extend timed automata (TA) by feature constraints to so-called featured timed automata (FTA) facilitating efficient verification of real-time properties for entire product lines in a single run. In this paper, we propose a novel modeling formalism, called configurable parametric timed automata (CoPTA), extending expressiveness of FTA by supporting freely configurable and therefore a-priori unbounded timing intervals for real-time constraints, which are defined as feature attributes in extended feature models with potentially infinite configuration spaces. We further describe an efficient test-suite generation methodology for CoPTA models, achieving location coverage on every possible model configuration. Finally, we present evaluation results gained from applying our tool implementation to a collection of case studies, demonstrating efficiency improvements compared to a variant-by-variant analysis.