A set of metrics for characterizing simulink model comprehension

Abstract

Simulink is a powerful tool for Embedded Systems, playing a key role in dynamic systems modeling. However, far too little attention has been paid to quality of Simulink models. In addition, no research has been found linking the relationship between model complexity and its impact in the comprehension quality of Simulink models. The aim of this paper is to define a set of metrics to support the characterization of Simulink models and to investigate their relationship with the model comprehension property. For this study, we performed a controlled experiment using two versions of a robotic Simulink model — one of them was constructed through the ad hoc development approach and the other one through the re-engineered development approach. The results of the experiment show that the re-engineered model is more comprehensible than the ad hoc model. In summary, the set of metrics collected from each version of the Simulink model suggests an inverse relationship with the model comprehension, i.e., the lower the metrics, the greater the model comprehension.