High Performance Real-Time Scheduling Framework for Multiprocessor Systems

Embedded systems, performing specific functions in modern devices, have become pervasive in today's technology landscape. As many of these systems are real-time systems, they necessitate operations with stringent time constraints. This is especially evident in sectors like automotive and aerospace. This thesis introduces a High Performance Real-time Scheduling (HPRTS) framework, which is designed to navigate the multifaceted challenges faced by multiprocessor real-time systems. To begin with, the research attempts to bridge the gap between system reliability and resource sharing in Mixed-Criticality Systems (MCS). In addressing this, a novel fault-tolerance solution is presented. Its main goal is to enhance fault management and reduce blocking time during fault tolerance. Following this, the thesis delves into task allocation in systems with shared resources. In this context, we introduce a distinct Resource Contention Model (RCM). Using this model as a foundation, our allocation strategy is formulated with the aim to reduce resource contention. Moreover, in light of the escalating system complexity where tasks are represented using Directed Acyclic Graph (DAG) models, the research unveils a new Response Time Analysis (RTA) for multi-DAG systems. This particular analysis has been tailored to provide a safe and more refined bound. Reflecting on the contributions made, the achievements of the thesis highlight the potency of the HPRTS framework in steering real-time embedded systems toward high performance.