I discussed with Paul several articles we wrote recently, namely his DATE’05 and DATE’06 papers, and my DSN’04 paper. They present new scheduling methods for reliable embedded systems. Two different fault-tolerance techniques are used: active redundancy (i.e., multiple execution of a task onto several processors) and passive redundancy (i.e., re-execution of a task after a failure). While the second one is dynamic, slack time is reserved statically in the schedule to obtain transparent fault-tolerance, and hence predictability. We also discussed methods to guarantee a desired level of reliability in embedded systems.

Réalisation Axome - Création de sites Internet