Todays enterprises rely entirely on their information systems, usually connected to the internet. Network access control, mainly ensured by firewalls, has become a paramount necessity. Still, the management of manually configured firewall rules is complex, error prone, and costly for large networks. The use of high abstract models such as role based access control RBAC has proved to be very efficient in the definition and management of access control policies. The recent interest in role mining which is the bottom-up approach for automatic RBAC configuration from the already deployed authorizations is likely to further promote the development of this model. Recently, an extension of RBAC adapted to the specificities of network access control, which we refer to as NS- RBAC model, has been proposed. However, no effort has been made to extend the bottom-up approach to configure this model. In this paper, we propose an extension of role mining techniques to facilitate the adoption of a model based framework in the management of network access control. We present policy mining, a bottom-up approach that extracts instances of the NS-RBAC model from the deployed rules on a firewall. We provide a generic algorithm that could adapt most of the existing role mining solutions to the NS-RBAC model. We illustrate the feasibility of our solution by experimentations on real and synthetic data.