Predictability and Efficiency in Contemporary Hard RTOS for Multiprocessor Systems

Hard real-time operating systems have been mostly designed for uniprocessors. Driven by the high computational demand of real-time systems, these operating systems have been re-designed for multiprocessors. Researchers have been identifying issues that affect the predictability of multiprocessor hard real-time systems. These sources of unpredictability make it difficult to accurately predict safe and tight worst-case execution time bounds of real-time tasks. As a result, the performance of multiprocessor hard real-time systems is compromised. In this paper, we present a middleware architecture that aims to provide predictability and high efficiency for multiprocessor hard real-time systems. The middleware supports hybrid hard real-time tasks, which are adaptive hard real-time services that intend to ensure predictability and performance. We also present a prototype of this architecture. Moreover, we identify and discuss issues that affect the predictability and performance of multiprocessor hard real-time systems.