Platform model composition framework for the development of real-time control systems

Abstract

Recent trend imposes stringent requirements on the design of embedded systems, differentiating them from general-purpose computer systems such as power consumption, timeliness, reliability, etc. Traditional platform models based on a single view of the underlying support software and hardware are not adequate for modeling all of the interference between the run-time tasks, resources, and services within a system caused by their behaviors and properties. This is mainly due to the fact that application software is tightly tuned to a particular platform, or it is designed and developed to be platform-specific. To overcome this limitation, a platform modeling framework for model-based software development is proposed by identifying the ranges of acceptable platform properties for application software and by specifying the models of computation with respect to the nonfunctional constraints on the underlying execution platform. Specifically, the focus is a multicore-based compositional platform model with fault tolerance for the developed framework. As a case study, a multicore real-time scheduling algorithm is applied to the framework, and the simulation results demonstrate the efficacy of the usability of the framework for supporting fault tolerance. Our proposed approach outperforms by 8.5% even for very heavy loaded system compared to the existing method. © 2005 - Ongoing JATIT & LLS.