A Survey-Based Empirical Evaluation of Bad Smells in LabVIEW Systems Models

Bad smells are indications of poor designs that decrease the quality and maintainability of software. Compared with extensive research on bad smells in the context of Object-Oriented Programming (code smells), bad smells in systems models (model smells) need much more investigation. Although some works have proposed several model smells in a few modeling domains, the understanding and perception of different types of model smells may vary due to depth of knowledge and area of expertise. To fill this gap, we conducted an empirical study to evaluate the model smells summarized in the existing literature within the context of LabVIEW systems models through an anonymous online survey. Based on the 45 complete responses received from a diverse group of systems modelers, we observed that there exist differences regarding the perception of various model smells. Furthermore, depth of knowledge (experienced and inexperienced users) was observed as a factor that affects a user's understanding of different model smells. However, area of expertise (academia/industry, as well as domain of focus) did not show significant difference in model smells perception. Moreover, we identified additional model smells from this empirical study. In this paper, we provide several recommendations to avoid common model smells and we summarize the lessons learned from this investigation. Our exploratory research provides empirical evidence that drives deeper insights into model smells and lays out recommendations to practitioners on how to avoid some of the prominent smells, thus improving the quality of software artifacts in systems models.