User-Centric Security and Privacy Approaches in Untrusted Environments
- Author(s): Diallo, Mamadou Hassimiou
- Advisor(s): Mehrotra, Sharad
- et al.
Internet based information systems (and applications) such as cloud and Internet of Things (IoT) systems have evolved significantly in recent years due to advances in computing infrastructure and networking technologies. Organizations of all types and sizes are eager to adopt these systems to manage their information. In fact, these systems are significantly impacting the way we work, socialize, create and share information, and organize the ow of people, ideas, and things around the world. As a result, there is a dramatic growth of the number of people who can access services provided by these systems.
While these systems provide significant benefits to users, service providers do not provide direct mechanisms for the users to control security and privacy of their data at the service providers. Existing practice requires users to implicitly trust the service providers and their security policies in managing and protecting users data. Loss of control over data as well as high security requirements of certain government agencies and organizations prevent endusers to fully trust service providers. The main reason being that existing cryptographic solutions for security and privacy are not reliable for rapidly changing data processing model used by IoT service providers. Cryptographic schemes are either deterministic (enables processing of encrypted data, but not very secure) or randomized (very secure, but doesn't support data processing). As a consequence, data is processed in the clear at the service providers, which lead to vulnerabilities for both insider and outsider attacks.
In this thesis, we particularly focus on user-centric, untrusted cloud and IoT based information systems, and investigate techniques and approaches to empower users to take control of their security and privacy needs in these untrusted environments. Cloud computing services are organized into three broad categories: Infrastructure as a Service (IaaS), Platform as a Service (PaaS), Software as a Service (SaaS). The same service models are also applicable to IoT services. We use these service models as the basis for performing deeper analysis of data security and privacy issues and challenges in both, cloud and IoT systems. Our proposed solutions span across these service models (IaaS, PaaS, SaaS) for both, cloud and IoT systems.
We first present Nomad, a framework for building highly secure cloud-based systems in the PaaS and IaaS models of cloud computing. The underlying cryptographic scheme used by Nomad to secure data in the cloud is a fully homomorphic encryption scheme, which has the advantage of enabling computations to be performed directly on encrypted data. Nomad employs a GPU-based parallelization mechanism to accelerate the expensive homomorphic encryption operations. We implemented an end-to-end, cloud-based, mission-critical defense application, using the Nomad framework to demonstrate its feasibility.
We then propose CloudProtect, a policy-based, extensible privacy middleware that empowers users to encrypt their sensitive data stored within various third-party cloud applications in the SaaS cloud model. It supplements existing cloud applications with a mechanism for protecting users data, with minimal disruption of user experience. We implemented a prototype of the privacy middleware with two applications, namely, Google Calendar and Google Docs. Our experimental results highlight the usefulness of the tradeoff between privacy, usability, and efficiency.
We finally introduce IoTtrust, a framework to ensure trust and con defense in IoT systems for both users and the infrastructure in the IaaS and PaaS models of IoT. IoTtrust includes an IoT policy language for specifying users' preferences and the system policies regarding users data collection and processing by the system; a contract management mechanism for generating application level contracts between users and the IoT system based on their agreed policies, and translating those contracts into the IoT devices, where they need to be enforced; and attestation schemes for attesting the integrity of contracts translation, and contracts enforcement across all devices. We implemented IoTtrust and integrated it with TIPPERS to make it a trustworthy IoT system.