Trust Establishment Mechanisms for Distributed Service Environments

The aim and motivation of this dissertation can be best described in one of the most important application fields, the cloud computing. It has changed entire business model of service-oriented computing environments in the last decade. Cloud computing enables information technology related services in a more dynamic and scalable way than before – more cost-effective than before due to the economy of scale and of sharing resources. These opportunities are too attractive for consumers to ignore in today’s highly competitive service environments. The way to realise these opportunities, however, is not free of obstacles. Services offered in cloud computing environments are often composed of multiple service components, which are hosted in distributed systems across the globe and managed by multiple parties. Potential consumers often feel that they lose the control over their data, due to the lack of transparent service specification and unclear security assurances in such environments. These issues encountered by the consumers boiled down to an unwillingness to depend on the service providers regarding the services they offer in the marketplaces. Therefore, consumers have to be put in a position where they can reliably assess the dependability of a service provider. At the same time, service providers have to be able to truthfully present the service-specific security capabilities. If both of these objectives can be achieved, consumers have a basis to make well-founded decisions about whether or not to depend on a particular service provider out of many alternatives. In this thesis, computational trust mechanisms are leveraged to assess the capabilities and evaluate the dependability of service providers. These mechanisms, in the end, potentially support consumers to establish trust on service providers in distributed service environments, e.g., cloud computing. In such environments, acceptable quality of the services can be maintained if the providers possess required capabilities regarding different service-specific attributes, e.g., security, performance, compliance. As services in these environments are often composed of multiple services, subsystems and components, evaluating trustworthiness of the service providers based on the service-specific attributes is non-trivial. In this vein, novel mechanisms are proposed for assessing and evaluating the trustworthiness of service providers considering the trustworthiness of composite services. The scientific contributions towards those novel mechanisms are summarised as follows: • Firstly, we introduce a list of service-specific attributes, QoS+ [HRM10, HHRM12], based on a systematic and comprehensive analysis of existing literatures in the field of cloud computing security and trust. • Secondly, a formal framework [SVRH11, RHMV11a, RHMV11b] is proposed to analyse the composite services along with their required service-specific attributes considering consumer requirements and represent them in simplified meaningful terms, i.e., Propositional Logic Terms (PLTs). • Thirdly, a novel trust evaluation framework CertainLogic [RHMV11a, RHMV11b, HRHM12a, HRHM12b] is proposed to evaluate the PLTs, i.e., capabilities of service providers. The framework provides computational operators to evaluate the PLTs, considering that uncertain and conflicting information are associated with each of the PLTs and those information can be derived from multiple sources. • Finally, harnessing these technical building blocks we present a novel trust management architecture [HRM11] for cloud computing marketplaces. The architecture is designed to support consumers in assessing and evaluating the trustworthiness of service providers based on the published information about their services. The novel contributions of this thesis are evaluated using proof-of-concept-system, prototype implementations and formal proofs. The proof-of-concept-system [HRMV13, HVM13a, HVM13b] is a realisation of the proposed architecture for trust management in cloud marketplaces. The realisation of the system is implemented based on a self-assessment framework, proposed by the Cloud Security Alliance, where the formal framework and computational operators of CertainLogic are applied. The realisation of the system enables consumers to evaluate the trustworthiness of service providers based on their published datasets in the CSA STAR. A number of experiments are conducted in different cloud computing scenarios leveraging the datasets in order to demonstrate the technical feasibility of the contributions made in this thesis. Additionally, the prototype implementations of CertainLogic framework provide means to demonstrate the characteristics of the computational operators by means of various examples. The formal framework as well as computational operators of CertainLogic are validated against desirable mathematical properties, which are supported by formal algebraic proofs.