Model Checking of Software Product Lines in Presence of Nondeterminism and Probabilities

Nowadays, Software Product Lines (SPLs) are being used in a variety of domains including safety-critical systems for which verification of the systems is a matter of concern. Formal modeling and verification of SPLs has been majorly investigated recently. Due to the potential large number of the products in a SPL, individual verification of all products could be costly or even impractical. Hence, there is a need for verification methods that can verify the whole family's behavior at once. In this paper, we focus on the probabilistic model checking of software product lines in which the behavior of individual products can be described in terms of Markov decision processes. We introduce a mathematical model, Markov Decision Process Family (MDPF), to compactly represent the behavior of the whole family. We also provide a model checking algorithm in order to verify MDPFs against properties expressed in probabilistic computational tree logic.