Optimal design of fault-tolerant soft-real-time systems with imprecise computations

Abstract

The imprecise computation technique applied to real-time systems has been proposed for a better control of the delivered service quality when full quality service cannot be achieved because of temporary overloads or reductions of computational power. We propose a methodology for the evaluation of the service quality delivered by a t’ault-tolerant multiprocessor soft-real-time system which employs the imprecise computation technique, subject to a periodic workload. Such a methodology allows us to define optimization strategies whose goal is to determine the optimal trade-off between the system cost and the overall quality of the delivered service.