Simulation-Based CTMC Model Checking: An Empirical Evaluation
Joost-Pieter Katoen, Ivan S. Zapreev · 2009
This paper provides an experimental study of the efficiency of simulation-based model-checking algorithms for continuous-time Markov chains by comparing: MRMC - the only tool that implements (new) confidence-interval-based algorithms for verification of all main CSL formulae; Ymer - that allows for verification oftime-bounded and time-interval until using sequential acceptance sampling; and VESTA - that can verify time-bounded and unbounded until by means of simple hypothesis testing. The study shows that MRMC provides the most accurate verification results. Ymer and VESTA, unlike MRMC, have almost constant memory consumption. Ymer requiresthe least number of observations to assess the model-checking problem, but MRMC is mostly the fastest. This indicates that the tools' efficiency does not so muchdepend on sampling but is rather determined by extra computations.