Abstract:

The paper starts with a brief overview on performance and reliability modelling paradigms for concurrent processes and their mapping to a Markov chain which is composed of marginal absorbing Markov chains. We use a new technique called compositional uniformization which allows for the evaluation of the mean time for a complete synchronization cycle which can be expressed as the mean time to absorption of concurrent and independent marginal Markov chains. The mean time to absorption exhibits a compositional character in the sense that quantities belonging to the marginal chains are combined to yield the desired cumulative measure. We demonstrate the efficiency of the technique by means of a very large example. Finally we discuss an extension to the field of reliability evaluation; we sketch how to calculate the exact MTTF for heterogeneous k-of-m systems by usage of the new technique.