Performance analysis of wireless sensor networks is a difficult task because of the high dynamic of networks and the use of duty-cycled MAC protocols. Markov-based modelling is an interesting approach to deal with this problem. However, existing Markov-based analytic models, being MAC protocol-centric rather than network-centric, work under strong assumptions and do not allow to encompassing important network parameters like radio channel fading and capture effect, or actual implementation optimizations (not always specified in the protocol description). In this paper we propose a novel approach to obtain a Markov chain model for networks running different MAC protocols by means of Process Mining Techniques. We present the main aspects of our approach together with the results obtained for the standard IEEE 802.15.4. The obtained Markov model can be used to evaluate various performance parameters. The approach can also be extended to a wider range of protocols.