Node-level Fault Tolerance for Fixed Priority Scheduling

Project no: 9811-4

Johan Karlsson

Chalmers University of Technology, Department of Computer Engineering

Application as pdf, ps. Complement to application as TEXT, pdf, support letter from Saab Ericsson Space AB.

Support: 1 PhD student for 2 years decided 98-12-09.

Start 99-04-01 with Joakim Aidemark as Phd student.

Reports:

• Updated project plan as pdf
• report 99-08-17
• First year report 00-06-15
• Second year report 02-01-10

 

Industry contacts

Torbjörn Hult torbjorn.hult@space.se

Saab Ericsson Space AB

Computer systems

Göteborg

Project: Node-level Fault Tolerance for Fixed Priority Scheduling

Overview

Distributed real-time systems are increasingly being used to control critical functions in automotive and aerospace applications. These systems must be fault-tolerant to be safe and reliable. The goal of this project is to develop techniques for node-level transient fault tolerance for distributed real-time systems using fixed priority scheduling. Specially a case study will be performed, where a small real-time kernel that achieves transient fault-tolerance will be implemented and validated using fault injection. The real-time kernel will be implemented for the Thor microprocessor developed by Saab Ericsson Space AB.

Results

Progress year 1

• Fault-tolerance techniques and fault-tolerant scheduling have been studied.
• Fault injection experiments have been performed on the Thor microprocessor using a quick-sort workload with different input sequences to estimate fault coverage and fault latency.
• A small real-time kernel has been developed for the Thor microprocessor.
• An extended abstract has been written and presented at the ARTES Graduate student conference March 16-17, 2000 at Chalmers.
• A fast abstract has been written and submitted to the International Conference on Dependable Systems and Networks in New York, USA, June 25-28, 2000.

Co-operation

The project is carried out in co-operation with Saab Ericsson Space AB. Saab Ericsson Space AB have provided a VHDL model of the Thor microprocessor and also a computer board with a physical Thor microprocessor. Two project meetings have been conducted together with Saab Ericsson Space AB during the year. There has also been several contacts during the development of the real-time kernel for the Thor microprocessor, where Saab Ericsson Space AB have provided valuable help. Further, we have given a presentation and demonstration of the MEFISTO fault injection tool to Saab Ericsson Space AB.

Publications

For a more up to date list please see Joakim Aidemarks page

1. Joakim Aidemark and Johan Karlsson. Using Massive Time Redundancy to Achieve Node-level Transient Fault Tolerance. ARTES Graduate Student Conference 2000.
2. J. Aidemark, P. Folkesson, and J. Karlsson, "Path-Based Error Coverage Prediction", in Journal of Electronic Testing Theory and Applications (JETTA), Vol. 16, June 2002, pp 343-349
3. J. Aidemark, J. Vinter, P. Folkesson, and J. Karlsson, "Experimental Evaluation of Time-redundant Execution for a Brake-by-wire Application", in Proc. International Conference on Dependable Systems and Networks (DSN-2002), Washington DC, USA, June 2002. pp 210-215.