Automated Generation and Analysis of Markov Reward Models Using Stochastic Reward Nets

Abstract

Markov and Markov reward models are widely used for the performance and reliability analysis of computer and communication systems. Models of real systems often contain thousands or even millions of states. We propose the use of Stochastic Reward Nets (SRNs) for the automatic generation of these large Markov reward models. SRNs do allow the concise specification of practical performance, reliability and per-formability models.An added advantage of using SRNs lies in the possibility of analyzing the (time-independent) logical behavior of the modeled system. This helps both the validation of the system (is the right system being built?) and of the model (does the model correctly represent the system?).We discuss the methods to convert SRNs into Markov reward processes automatically. We review the solution techniques for the steady state and transient analysis of SRNs and Markov reward processes. We also discuss methods for the sensitivity analysis of SRNs.KeywordsFiring TimeReachability GraphInput PlaceReward RateFiring SequenceThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.