|Title:||Reliability framework for distributed real-time systems|
Redundancy in the physical, temporal, information and analytical domains is the key for achieving fault tolerance and due to the wealth of research in this domain, a rich set of techniques has been successfully used in many critical applications. The most common ways to enable redundancy in real-time systems task are time redundancy, i.e., re-executions of recovery blocks, and space redundancy, e.g., N-modular redundancy. This project will enable the combination of the two redundancy approaches in real-time systems consisting of tasks with mixed criticalities.
The goal of this thesis is to implement a framework for space and time redundancy in distributed real-time systems. It will provide platform to:
a) Derive scheduler specific attributes for tasks with various criticalities scheduled on networked real-time systems
b) Evaluate the schedulability of different (existing) task sets in different schedulers when the criticalities change
1. Using optimization theory
2. Addressing distribution aspects on, e.g., CAN and FlexRay