Feature Interaction in Pervasive Computing Systems

Abstract

Feature interaction describes a situation where the combination of two or more services that individually perform correctly results in unexpected and possibly adverse behaviour. Such feature interaction issues have first been identified in telecommunication systems and are now beginning to be considered in other distributed software systems. We expect significant feature interaction research in pervasive computing where very many applications collaborate and adapt to changes to their environment or to user needs in order to provide tailored services to users. This paper presents a classification of feature interaction issues in pervasive computing systems. The classification captures, with a focus on automotive systems and systems for smart homes, feature interaction issues related to types of interaction, channels of interaction, and user needs. The classification aims to aid the understanding of feature interaction in pervasive computing systems, and to serve as a guideline for designers of pervasive applications.