Continuous trust management frameworks : concept, design and characteristics

Abstract

A Trust Management Framework is a collection of technical components and governing rules and contracts to establish secure, confidential, and Trustworthy transactions among the Trust Stakeholders whether they are Users, Service Providers, or Legal Authorities. Despite the presence of many Trust Frameworks projects, they still fail at presenting a mature Framework that can be Trusted by all its Stakeholders. Particularly speaking, most of the current research focus on the Security aspects that may satisfy some Stakeholders but ignore other vital Trust Properties like Privacy, Legal Authority Enforcement, Practicality, and Customizability. This thesis is all about understanding and utilising the state of the art technologies of Trust Management to come up with a Trust Management Framework that could be Trusted by all its Stakeholders by providing a Continuous Data Control where the exchanged data would be handled in a Trustworthy manner before and after the data release from one party to another. For that we call it: Continuous Trust Management Framework. In this thesis, we present a literature survey where we illustrate the general picture of the current research main categorise as well as the main Trust Stakeholders, Trust Challenges, and Trust Requirements. We picked few samples representing each of the main categorise in the literature of Trust Management Frameworks for detailed comparison to understand the strengths and weaknesses of those categorise. Showing that the current Trust Management Frameworks are focusing on fulfilling most of the Trust Attributes needed by the Trust Stakeholders except for the Continuous Data Control Attribute, we argued for the vitality of our proposed generic design of the Continuous Trust Management Framework. To demonstrate our Design practicality, we present a prototype implementing its basic Stakeholders like the Users, Service Providers, Identity Provider, and Auditor on top of the OpenID Connect protocol. The sample use-case of our prototype is to protect the Users’ email addresses. That is, Users would ask for their emails not to be iii shared with third parties but some Providers would act maliciously and share these emails with third parties who would, in turn, send spam emails to the victim Users. While the prototype Auditor would be able to protect and track data before their release to the Service Providers, it would not be able to enforce the data access policy after release. We later generalise our sample use-case to cover various Mass Active Attacks on Users’ Credentials like, for example, using stolen credit cards or illegally impersonating third-party identity. To protect the Users’ Credentials after release, we introduce a set of theories and building blocks to aid our Continuous Trust Framework’s Auditor that would act as the Trust Enforcement point. These theories rely primarily on analysing the data logs recorded by our prototype prior to releasing the data. To test our theories, we present a Simulation Model of the Auditor to optimise its parameters. During some of our Simulation Stages, we assumed the availability of a Data Governance Unit, DGU, that would provide hardware roots of Trust. This DGU is to be installed in the Service Providers’ server-side to govern how they handle the Users’ data. The final simulation results include a set of different Defensive Strategies’ Flavours that could be utilized by the Auditor depending on the environment where it operates. This thesis concludes with the fact that utilising Hard Trust Measures such as DGU without effective Defensive Strategies may not provide the ultimate Trust solution. That is especially true at the bootstrapping phase where Service Providers would be reluctant to adopt a restrictive technology like our proposed DGU. Nevertheless, even in the absence of the DGU technology now, deploying the developed Defensive Strategies’ Flavours that do not rely on DGU would still provide significant improvements in terms of enforcing Trust even after data release compared to the currently widely deployed Strategy: doing nothing!