FaultTolerance for Distributed and RealTime Embedded Systems Ostre

Fault-Tolerance for Distributed and Real-Time Embedded Systems Ostre Team Yves Sorel Bip Team Alain Girault Hamoudi Kalla Workshop De. FINE : Pisa - Italy November 26 -27, 2002

Outline ¢ Introduction: Real-Time Embedded System ¢ Fault-Tolerance: Our Methodology ¢ Proposed Methods ¢ Conclusion and Future Work

Introduction: real-time embedded system Failures models Sensors Actuators Architecture : processors and communications components Applications : C functions, Caml …. Syn. DEx tool implementation of real-time embedded applications on multicomponent architectures

Fault-Tolerance: Our Methodology High level program Compiler Architecture specification Model of the algorithm Distribution constraints Execution times Distribution heuristic Real-time constraints Failure specification Fault-tolerant distributed static schedule Code generator Fault-tolerant distributed embeded code

The Proposed Solutions: Models P 1 B D in A C out P 2 P 3 D Algorithm Model Architecture Model P 4

The Proposed Solutions: (1) P 1 Schedule Architecture Model Schedule 1 P 2 P 3 P 4 in Failure Model 1 P 1 B A Schedule 2 P 3 P 4 out Failure model 2 Algorithm Model P 1 P 2 Schedule 3 P 3 Failures Models P 4 Fault-tolerant distributed static schedule

The Proposed Solutions: (2) B P 1 in P 2 L. 1 -2 -3 P 3 L. 3 -4 Architecture in in Model send P 2 B A P 3 P 4 Schedule send N = 2 out Algorithm N : Number of processor failures Model Failures Models P 4 Fault-tolerant distributed static schedule

Conclusion & Future Work ¢ Develop a new fault-tolerance method, we take into account : l Reliability Measures which is the probability that a component will not fail during a certain amount of time. l Both processors and links failures.

References ¢ Catalin Dima, Alain Girault, Christophe Lavarenne, and Yves Sorel. Off-Line Real-Time Fault. Tolerant Scheduling. Euromicro Workshop on Parallel and Distributed Processing. Mantova, Italy, February 2001. ¢ Alain Girault, Christophe Lavarenne, Mihaela Sighireanu, and Yves Sorel. Fault-Tolerant Static Scheduling for Real-Time Distributed Embedded Systems. 21 st International Conference on Distributed Computing Systems. Phoenix, USA, April 2001. ¢ Alain Girault, Christophe Lavarenne, Mihaela Sighireanu, and Yves Sorel. Generation of Fault. Tolerant Static Scheduling for Real-Time Distributed Embedded Systems with Multi-Point Link. IEEE Workshop on Fault-Tolerant Parallel and Distributed Systems. San Francisco, USA, April 2001.