Characterization of Implied Scenarios as Families of Common Behavior

Abstract Concurrent systems face a threat to their reliability in emergent behaviors, which are not included in the specification but can happen during runtime. When concurrent systems are modeled in a scenario-based manner, it is possible to detect emergent behaviors as implied scenarios (ISs) which, analogously, are unexpected scenarios that can happen due to the concurrent nature of the system. Until now, the process of dealing with ISs can demand significant time and effort from the user, as they are detected and dealt with in a one by one basis. In this paper, a new methodology is proposed to deal with various ISs at a time, by finding Common Behaviors (CBs) among them. Additionally, we propose a novel way to group CBs into families utilizing a clustering technique using the Smith-Waterman algorithm as a similarity measure. Thus allowing the removal of multiple ISs with a single fix, decreasing the time and effort required to achieve higher system reliability. A total of 1798 ISs were collected across seven case studies, from which 14 families of CBs were defined. Consequently, only 14 constraints were needed to resolve all collected ISs, applying our approach. These results support the validity and effectiveness of our methodology.