Applying The Semantic Web Layers To Access Control
Applying The Semantic Web Layers To Access Control
Applying The Semantic Web Layers To Access Control
Mariemma I. Yag ue, Antonio Ma na, Javier L opez, Jos e M. Troya Computer Science Department, University of Malaga, Spain {yague, amg, jlm, troya}@lcc.uma.es
Abstract
The Semantic Web, also known as the Web of meaning, is considered the new generation of the Web. Its objective is to enable computers and people to work in cooperation. A requisite for this is encoding data in forms that make web contents (meaning, semantics) more understandable by algorithmic means. In this paper, we present the application of Semantic Web concepts and technologies to the access control area. The Semantic Access Control Model (SAC) uses different layers of metadata to take advantage of the semantics of the different components relevant for the access decision. We have developed a practical application of this access control model based on a specic language, denominated Semantic Policy Language (SPL), for the description of access criteria. This work demonstrates how the semantic web concepts and its layers infrastructure may play an important role in many relevant elds, such as the case of access control and authorization elds.
Specically, we are concerned with the semantic integration of external authorization entities in the applications and to provide a distributed and scalable framework supporting advanced authorization and access control schemes in an efcient way. In this paper we present an access control model that addresses the aforementioned problems. This model is based on the use of semantic descriptions of the authorization entities; separation of the attribute certication and the authorization management functions, following the layers infrastructure of the Semantic Web. Section 2 states the fundamentals of the access control problem. Section 3 highlights the basis of semantic modelling through XML metadata. The fundamentals of the Semantic Access Control Model are presented in section 4, along with a detailed description of its semantic layers. Finally, section 5 underlines some concluding results and future research lines.
1 Introduction
The problem of semantic interoperation has been extensively studied. This problem appears when different applications mean different things by similar terms. Semantic heterogeneity is closely tied to the context-dependent interpretations of the concepts represented. Today, interoperability appears as the main challenge to address concerning authorization for open and distributed systems. Access control is a critical component in many environments. Access to systems and resources has to be controlled in a safe way. Nowadays, security issues for semantic web services are becoming more important. This is the reason the Semantic Web has included the access control area among the objectives of its Advanced Development (SAWD) projects. Our work is based on the denition of semantic models upon the components of an access control model to reach interoperability through the semantic integration in heterogeneous and distributed environments.
Work
the privileges is stored and managed locally in the same system where the resources reside. The most relevant problem that this scheme presents when applied to open distributed systems is the lack of interoperability. It is not reasonable to expect that heterogeneous systems for different purposes and under control of different parties will be able to dene a common homogeneous set of authorization criteria. Lets review some of the characteristics of these new systems and applications that are relevant to the design of the access control model. Heterogeneity. First of all, it is important to keep in mind that, in open distributed systems it is frequent to have a large number of stakeholders or owners of resources with very different policies and interests. A large number of previously unknown clients that are impossible to classify in advance is also predictable. Moreover, resources found in distributed systems are intrinsically heterogeneous. Heterogeneity affects not only to the type of resource but also to the format, origin, validity, etc. This heterogeneity of resources, clients and owners implies very disparate security requirements and access control criteria. Interoperability. It has been mentioned that the authorization approach in most of current systems relies on locally-issued credentials related to user identity. This type of credentials presents many drawbacks, but the most important is that they are not interoperable. Taking into account security, scalability and interoperability, the separation of the certication of attributes and access control management responsibilities is widely accepted as a scalable and exible solution. The external authorization infrastructure is known as PMI. The main entities of a PMI, known as Source of Authorizations (SOAs), issue attribute certicates. Access control systems select which SOAs to trust and which combination of attributes to use as access criteria. To achieve interoperability, a mechanism to convey the semantics of the attributes certied by the SOAs to the access control systems is required. Flexibility. Due to the heterogeneity and taking into account that our model is designed to be applied in open distributed systems where numerous specic systems will coexist and interoperate, exibility appears as one of the most important goals to achieve. The model must be exible enough to be applicable in different scenarios with few or no changes. o Scalability. In the systems that we are considering we deal with very large numbers of resources, access policies, systems, clients and attributes. Therefore, the scalability of the scheme is very important. To achieve this scalability a fully distributed scheme is mandatory. Dynamism. The access control model must be capable of adapting itself to frequent changes in different parameters such as access criteria, client attributes, environment conditions, resources available, etc. To avoid management overload due to the control of changes, the model must adapt in
a transparent and automatic way to these changes. The basic concepts upon which the access control model is based determine the exibility of the model to adapt to different environments and systems. Several access control models have been developed based on different schemes. It is important to realize that the existing access control models were developed for closed environments. Consequently, they are built on the basis of modeling the environments that motivated their development. Lets review these models. Discretionary Access Control (DAC) was designed for multi-user databases and systems with a few, previously known, users. Changes were rare and all resources were under control of a single entity. Access controlled based on the identity of the requestor and on access rules stating what requestors are (or are not) allowed to do [2]. Mandatory Access Control (MAC) had its origins in military environments where the number of users can be quite high, but with a static, linearly hierarchic classication of these users. The model is based on the denition of a series of security levels and the assignment of levels to resources and users. MAC policies control access based on mandated regulations determined by a central authority [6]. Role-based Access Control (RBAC) is inspired in the business world. The development of RBAC coincides with the advent of corporate intranets. Corporations are usually hierarchically structured and access permissions depend on the position of the user in the hierarchy, i.e. the role played by the user. RBAC policies control access depending on the roles that users play within the system and on rules stating what accesses are allowed to users in given roles [1]. Among the previous models RBAC is commonly considered a mature and exible technology. Consequently, it is the most popular paradigm in use today. The main problem with role based access control is that the mechanisms are built on three predened concepts: user, role and group. The denition of roles and the grouping of users can facilitate management, specially in corporate information systems, because roles and groups t naturally in the organizational structures of the companies. However, when applied to some new and more general access control scenarios, these concepts are somewhat articial. A more general approach is needed in these new environments. Groups are a specic use of a more general tool: the attribute. Groups are usually dened based on the values of attributes (position, ..). Other attributes, such as identity, are even built into most of the access control models. The identity is a useful attribute, but it should not be a builtin component of a general model. The static grouping of users of RBAC can sufce in corporate systems, but it is not exible enough to cope with the requirements of more dynamic environments where the structure of groups can not be foreseen by the administrators of the access control system. In these scenarios each resource may possibly need
a different group structure and access control policy. New resources are incorporated to the system continuously and policies for a given resource may change frequently. By considering attributes to be the basis of the access control model we can develop a very exible and open model that is able to be used in most scenarios. In fact, MAC, DAC and RBAC schemes can be specied using the attribute-based approach. In [5] we proposed a modular an dynamic approach based on the separation of the access control criteria from the rules of allocation of policies to resources. We called this scheme Dynamic Access Control. The new model that we present is called Semantic Access Control (SAC) because it complements the use of attributes as the building block of the model with the use of metadata to represent the semantics of the different elements.
of specifying access control criteria. The modular denition of SPL policies implies the separation of specication in three parts; that is, access control criteria, allocation of policies to resources and semantic information (properties about resources and context). Additionally, SPL makes possible the abstraction of access control components and, as a consequence, the ability to reuse these access control components. All the previous properties help the reduction of the complexity of management. Moreover, the use of semantic information about the context allows the administrator to include relevant contextual considerations in a transparent manner, also helping the semantic validation task.
mantic interoperability [10]. The SOAD metadata model has been applied to the semantic integration of an infrastructure of authorization entities in the CORBA architecture [4]. The SOAD metadata model along the semantic validation algorithms is extensively studied in [9]. In conclusion, the semantic approach of SAC is the foundation to achieve semantic interoperability among the different components of access control systems. SAC reaches the highest layers of the Semantic Web (Logical, Inference and Trust). The SOAD metadata model describes inference rules for deducting new information (incompatible, equivalent or derived attribute certicates) and supports the semantic validation of policies, providing proofs of the correctness of the access control policies. Additionally, the SOAD metamodel enables trusted interoperation between access control systems and external authorization entities. We are also working on the development of additional semantic models to enable secure delegation of attribute certicates. The extension of the Semantic Policy Language with additional digital rights specication, along with semantic models for its management is under consideration.
References
[1] D. Ferraiolo and D. Kuhn. Role based access control. In 15th NIST-NSA National Computer Security Conference, 1992. [2] B. W. Lampson. Protection. Computer Networks, 8(1):18 24, 1974. [3] O. Lassila and R. Swick. Resource description framework (rdf). Technical Report W3C Recommendation 1999-02-22, W3C, 1999. [4] J. L opez, A. Ma na, J. Ortega, E. Pimentel, J. Troya, and M. Yag ue. Integrating pmi services in corba applications. Computer Standards and Interfaces, 25(4):391409, 2003. [5] J. L opez, A. Ma na, and M. Yag ue. Xml-based distributed access control system. In EC-Web02, volume 2455 of LNCS. Springer-Verlag, 2002. [6] X. Qian and T. Lunt. A mac policy framework for multilevel relational databases. IEEE Transactions on Knowledge and Data Engineering, 8(1):114, 1996. [7] P. Samarati and S. de Capitani di Vimercati. Access control: Policies, models, and mechanisms. In FOSAD 2000, volume 2171 of LNCS, pages 137196. Springer-Verlag, 2001. [8] T. Woo and S. Lam. Designing a distributed authorization service. In Proc. of IEEE INFOCOM, volume 2437 of LNCS, pages 227245. Springer-Verlag, 1998. ue, A. Ma na, J. L opez, E. Pimentel, and J. Troya. [9] M. Yag A secure solution for commercial digital libraries. Online Information Review, 2003(3), 2003. [10] M. Yag ue and J. Troya. Semantic approach for access control in web services. In Euroweb02, eWiC. British Computer Society, 2002.