Behavioral Complexity Quantification (Becom-Q)

Abstract

Todays' embedded and cyber-physical systems (CPS) rely heavily on complex software functions. While part of this complexity is unavoidable and caused by a growing number of functions (intrinsic complexity), another part is related to inappropriate design and development methods (called avoidable complexity). Indeed the early removal of avoidable complexity is one of the key challenges of software-intensive systems.Over the years, several methods have been proposed to identify and quantify software complexity. However, current complexity measures rely mostly on the analysis of software structures and are not good predictors of software complexity and potential design flaws. This motivates the needs of new metrics of software complexity. In this paper, we describe a new complexity metric based on the analysis of behavioral aspects of software we call Behavioral Analysis Quantication or Becom-Q for short.Specically, we use advances in model-checking and model counting to quantify all possible behaviors that a software function may exhibit and use this quantication to model software complexity.