A Comparative Evaluation of UML Profile for Hard and Weakly Hard Real-Time Systems

Embedded systems that have timing constraints are classified as embedded real-time systems. Formally, embedded real-time systems are classified into three categories of timing constraints, named hard, soft and weakly hard. For hard real-time systems, no deadlines miss is tolerated, while in soft real-time systems, deadlines miss is acceptable occasionally. Meanwhile, for weakly hard real-time systems, a missed deadline is tolerated but has to be stated precisely. By far, modeling timing constraints and scheduling behavior through adaptation of modeling language are used to predict the timing behavior and performance of sets concurrent tasks in embedded real-time systems. Nowadays, the widely used, well-known and most adopted modeling language for software modeling is Unified Modeling Language (UML). In this paper, we studied the UML profile for Schedulability, Performance and Time (SPT) as modeling language because it is known as suitable profile for modeling and analysis of embedded real-time systems. The objectives of this paper are to compare the capability of UML-SPT profile and to propose modification UML-SPT metamodel to support both of hard real-time and weakly hard real-time requirements. Two case studies of embedded real-time systems used in this evaluation were: an elevator system case study which was chosen to represent a hard embedded real-time system and a mobile robot case study which was chosen to represent a weakly hard embedded real-time system. A set of important criteria to enable the modeling of timing requirements and scheduling analysis of embedded real-time systems were identified to conduct the comparative evaluation. The analysis results of the comparative evaluation showed how UML-SPT profile can support the requirements modeling and analysis of hard real-time systems and weakly hard real-time systems. A set of modification to improve UML-SPT metamodel in supporting the timing requirements was proposed in this paper. The modifications were applied in the case studies to show how the modification used in timing modeling and scheduling analysis of embedded real-time systems.