PERFORMANCE DEPENDABILITY MODELING USING STOCHASTIC REWARD MODELS - TRANSIENT-BEHAVIOR/

Stochastic Reward Models (SRMs) are commonly used for evaluating combi ned performance and dependability of fault-tolerant systems. An SRM is composed of a stochastic process, describing the evolution of the sys tem, and a superimposed reward structure, reflecting different perform ance levels of the system. Evaluating combined transient performance/d ependability measures using SRMs leads either to differential or integ ral equations. This paper discusses the use of SRMs for modelling perf ormance/dependability of fault-tolerant systems and proposes an approa ch for the numerical solution of integral equations which commonly ari se. A bound for the error due to the numerical approximation is obtain ed. As an example, an ii-unit parallel system is analysed numerically in transient stale. (C) 1998 Elsevier Science Ltd. All rights reserved .