Secure Deduplication Systems with Advanced Applications


Student thesis: Doctoral Thesis

View graph of relations


Related Research Unit(s)


Awarding Institution
Award date22 Aug 2018


Digital data are explosively generated nowadays, which brings tremendous pressure to the whole Internet. To cope with this challenge, storing data in globally distributed storage servers (or other advanced cache-enabled network devices) and applying data deduplication techniques are common solutions. However, such a wide attacking surface and many recent data breaches raise concerns about user privacy exposure and unauthorized data access. As simply performing encryption will inactivate computations on top of data, the benefits of data utilization like keyword search would diminish.

This dissertation presents algorithms and implementation for secure deduplication system and other related applications over encrypted storage, ensuring strong protection on data privacy while addressing the specific challenges in different application scenarios. First, we proposed a secure client-side deduplication middleware system, aiming for bringing the benefits of storage and bandwidth savings back to the user client while preserving their data privacy. Second, we designed a secure and accurate near-duplicate detection service, aiming for locating encrypted near-duplicates for authorized users from in-network data storage owned by multiple content providers. Third, we leveraged the correlation of outsourced image datasets and devised a secure and efficient cloud-assisted data sharing architecture for mobile devices with privacy assurance. The line of research has impacts on both academy and industry. The proposed solutions promote new algorithms and new architectures for applications built on encrypted storage with security and performance guarantees, which facilitate achieving grip among stakeholders across deduplicated storage, content-centric services, and so on.