Abstract-We propose an access control model that extends RBAC (Role-Based Access Control) to take time and location into account, and use term rewriting systems to specify access control policies in this model. We discuss implementation techniques for rewrite-based policy specifications, and the integration of these policies in web applications. The declarative nature of the model facilitates the analysis of policies and the evaluation of access requests: we present two case-studies.
I. INTRODUCTIONThe recent growth of digital communication has increased the demand of security for protecting resources and preserving the integrity and confidentiality of data. Thus, considerable interest has been centred on the area of access control models in recent years. Access control is concerned with deciding which actions a subject can execute on the objects of a given system. One of the most popular choices, for use with centralised systems, is the Role-Based Access Control (RBAC) model [19]. In RBAC, users are assigned to roles by a security administrator; roles usually map to job titles in an organisation, and as such, this model is well-suited for relatively static environments.High mobility of users and services in the emerging mobile applications entails the need for access control models that take the location of the user and the time of the request into account in order to decide whether to grant or deny an access request. Several extensions to RBAC have been proposed to incorporate spatio-temporal information in the model. One of the first time-based RBAC models was proposed in [5], and later generalised in the GTRBAC model [17]. In these systems, the roles are enabled by time constraints. An example of location-based model is introduced in [13]. The GEO-RBAC model allows a user to activate a role from a particular location, and the role and its permissions are predefined for that location. Other models considering both location and time constraints have been proposed, see for example [12], [11], [22], [9].In this paper we describe a location-and time-based RBAC model, TLRBAC, that we specify using term rewriting. Term