Resolving uncertainty in automotive feature interactions

Towards feature-oriented requirements validation for automotive systems

“…This method starts with an empty specification, and then incrementally adds clauses to the specification until all the feature requirements are satisfied. Arora et al [7] propose a method and algorithms for identifying and resolving feature interactions in the early stages of the software development life-cycle. The work uses State Machines to model the behavior of independent features, context diagrams to integrate independent features, and Live Sequence Charts to capture the interactions of features.…”

“…The validity of a feature set refers to two situations: 1) one is from the structural perspective, i.e., the selected features should conform to the constraints defined by the feature model and, 2) the other is from the functional perspective, i.e., no undesirable behaviors exist between two or more (as integrated) feature behavioral specifications. In order to increase the confidence of the validity of the feature set, several feature-oriented requirements validation techniques [3], [4], [5], [6], [7], [8] have been developed. However, it is well recognized that with the increasing size of feature models, the inherent variability of the feature sets leads to an inevitable issue that the hidden flaws of features are difficult to avoid [9].…”

“…In the literature, there are many examples [7], [10], [11] where the process starts with translating the natural language specification (NLS) of a feature into a formal language specification (FLS) of the feature behaviors, and then the requirements validation is performed based on the generated formalisms. To our best knowledge, the main drawbacks of using NLS lie in: 1) ambiguities in the NLS cause imprecise definitions and even wrong understanding of the feature behaviors and, 2) the direct translation from the NLS to a FLS tends to be very costly and, 3) the NLS hinders to a large extent the possibility of performing automatic feature-oriented requirements validation.…”

Uncertainty handling in goal-driven self-optimization – Limiting the negative effect on adaptation

“…This study handles requirements uncertainty at the language level where ours is at the model level. Arora et al (2012) focused on the uncertainty arising from inconsistent feature interactions. The difference is that our approach has an aim to improve or maintain the satisfaction level of requirements even though initially the requirements goal models are not perfectly defined.…”

“…While design-time uncertainty handling in requirements has been widely addressed through better approaches in elicitation, disambiguation and inconsistency check (Liaskos et al, 2012;Yang et al, 2012;Arora et al, 2012), these approaches more or less involve human intervention. Hence, runtime adaptations through unsupervised feedback loops present little time to apply them, while quickly responding to changes in the environments.…”