Using Software Architecture Descriptions to Detect Architectural Smells at Design Time

Abstract

Architectural smells are decisions made at the software architecture level, whether intentional or not, that may negatively impact the quality of a software system. In the literature, architectural smells are identified mainly by relying on the source code or other implementation artifacts. However, architectural smells could be detected at design time, even before employing implementation efforts and preventing them from being reflected at the system implementation. This research investigates how software architecture descriptions realized through architecture description languages (ADLs) can be used to identify architectural smells at design time. This work focuses on how architectural smells manifest and can be detected in SysADL, an ADL that allows describing both structure and behavior of software architectures using standardized diagrams from the OMG’s SysML language.