Theodosis Dimitrakos

On a generalized modularization theorem

The relation between a metalogical property of entailment (interpolation) and a structural property of categories of theory presentations (stability of faithful morphisms under pushouts) is studied in a general logical framework. The stability of faithful morphisms for any specific instance of the pushout construction is shown to be equivalent to the existence of interpolants over the corresponding locus of the underlying logic. The latter neither assumes nor requires interpolation to be a global property of the formalism.

Interpolation in practical formal development

Interpolation (together with completeness and decidability) has become one of the standard properties that logicians investigate when designing a logic. In this paper, we provide strong evidence that the presence of interpolants is not only cogent for scientific reasoning but has also important practical implications in computer science. We illustrate that interpolation in general, and uniform splitting interpolants, in particular, play an important role in applications where formality and modularity are invoked. In recognition of the fact that common logical formalisms often lack uniform interpolants, we advocate the need for developing general methods to (re)engineer a specification logic so that (at least) some critical uniform interpolants become available.

Policy-driven access control over a distributed firewall architecture

Motivated by a scientific application, where virtual organisations are dynamically created to achieve specific goals by sharing resources and information, we propose the synthesis of two lines of research: policy-based access control and distributed firewalls. Through this fusion we expect to deliver a scalable method of setting up security infrastructures for Grid computing infrastructures.

The coras approach for model-based risk management applied to e-commerce domain

The CORAS project develops a practical framework for model-based risk management of security critical systems by exploiting the synthesis of risk analysis methods with semiformal specification methods, supported by an adaptable tool-integration platform. The framework is also accompanied by the CORAS process, which is a systems development process based on the integration of RUP and a standardised security risk management process, and it is supported by an XML-based tool-integration platform. The CORAS framework and process are being validated in extensive user trials in the areas of e-commerce and telemedicine. This paper presents an overview of the CORAS framework, emphasising on the modelling approach followed in the first of the user trials (concerning the authentication mechanism of an e-commerce platform) and it provides some examples of the risk analyses employed in this context.

Towards Dynamic Security Perimeters for Virtual Collaborative Networks

Rapid technological advancements capitalising on the convergence of information (middleware) and communication (network) technologies now enable open application-to-application communication and bring about the prospect of ad hoc integration of systems across organisational boundaries to support collaborations that may last for a single transaction or evolve dynamically over a longer period. Architectures for managing networks of collaborating peers in such environments face new security and trust management challenges. In this paper we will introduce the basic elements of such an architecture emphasising trust establishment, secure collaboration, distributed monitoring and performance assessment issues.

Virtualised Trusted Computing Platform for Adaptive Security Enforcement of Web Services Interactions

Security enforcement framework is an important aspect of any distributed system. With new requirements imposed by SOA-based business models, adaptive security enforcement on the application level becomes even more important. Our work on the enforcement framework to date has resulted in a comprehensive middleware-based solution leveraging on Web services technologies. However, potential merits of hardware-based solutions to further secure application exposure have not been considered so far. This paper describes a method for combining software resource level security features offered by Web services technologies, with the hardware-based security mechanisms offered by trusted computing platform and system virtualisation approaches. In particular, we propose trust-based architecture for protecting the enforcement middleware deployed at the policy enforcement endpoints of Web and grid services. The main motivation is to additionally secure execution environment of the applications, by providing virtual machine level separation that maps from logical domains imposed by Web services level enforcement policies.

Compositional Structuring in the B-Method: A Logical Viewpoint of the Static Context

The B-Method provides a collection of structuring mechanisms which support information hiding, modularisation and compositionality of module operations, although, in order to achieve compositionality and independent (parallel) refinement, sharing is restricted in B. In this paper we elaborate some non-interference and compositionality assumptions that underlie structuring mechanisms such as USES, SEES and imports and show how they may be violated by inducing emerging properties which alter the context of the used, seen or imported machine. We discuss how such situations can be avoided by considering necessary and sufficient conditions for logical conservativeness and modularisation. As proof obligations, these conditions ensure that the properties of the context of the seen, used or imported component are conserved, i.e. that they are preserved but not enriched. From a logical viewpoint, these proof obligations require that the uniform interpolant of the contextual extension axioms is implied by the base context.

The CORAS Framework for a Model-Based Risk Management Process

CORAS is a research and technological development project under the Information Society Technologies (IST) Programme (Commission of the European Communities, Directorate-General Information Society). One of the main objectives of CORAS is to develop a practical framework, exploiting methods for risk analysis, semiformal methods for object-oriented modelling, and computerised tools, for a precise, unambiguous, and efficient risk assessment of security critical systems. This paper presents the CORAS framework and the related conclusions from the CORAS project so far.

Deploying Trust Policies on the Semantic Web

Automated services over open distributed systems have been advocated as the future for conducting business. However, in order for such services to be successfully deployed, confidence in the system and its participants needs to established. The use of trust policies and trust management has been proposed to raise the level of trust whilst still maintaining a practical level of automation. To define, deploy and enforce trust policies a level of common vocabulary and understanding between participants needs to be established. The Semantic Web initiative of the W3C has been developing common languages and tools to allow the exchange and processing of common vocabularies, in the form of a graph-based description format, and an ontology language. In this paper we discuss the steps required to establish a policy-based service architecture, discuss the role of the Semantic Web initiative in enabling the practical deployment of this architecture, and highlight the work required to enable this.

Towards dynamic population management of abstract machines in the B method

We study some restrictions associated with the mechanisms for structuring and modularising specifications in the B abstract machine notation. We propose an extension of the language that allows one to specify machines whose constituent modules (other abstract machines) may change dynamically, i.e., at run time. In this way, we increase the expressiveness of B by adding support for a common activity of the current systems design practice. The extensions were made without having to make considerable changes in the semantics of standard B. We provide some examples to show the increased expressive power, and argue that our proposed extensions respect the methodological principles of the B method.