Predict malfunction-prone modules for embedded system using software metrics

High software dependability is significant for many software systems, especially in embedded system. Dependability is usually measured from the user's viewpoint in terms of time between failures, according to an operational profile. A software malfunction is defined as a defect in an executable software product that may cause a failure. Thus, malfunctions are attributed to the software modules that cause failures. Developers tend to focus on malfunctions, because they are closely related to the amount of rework necessary to prevent future failures. This paper defined a software module malfunction-prone by class cohesion metrics when there is a high risk that malfunctions will be discovered during operations. Also proposed a novel cohesion measure method for derived classes in embedded system.