Home

Welcome!

Welcome to Data Security and Privacy Lab (DSP-Lab, founded in 2017) in the Department of Computer Science and Engineering at the University of Notre Dame. Here at DSP-Lab, we aim to explore and resolve security and privacy issues existing in the world of data so that people will be less involved in security or privacy breach in the big data era. The research areas of DSP-Lab at Notre Dame spans widely across multiple areas, from data science research to traditional security research, owing to the interdisciplinary nature of the security and privacy research in the big data.


Current and Past research projects

The following sections describe our current on-going research projects. Please contact Prof. Taeho Jung if you want further discussion.

Secure and verifiable COVID-19 tracing and containment

To mitigate the spread of COVID-19, public health authorities (PHAs) need to utilize individuals’ geo-spatio-temporal data, but such data are extremely sensitive because they contain individuals’ private daily footprints. We study how existing cryptographic primitives and trusted execution environment can be combined to build a secure, verifiable, and efficient framework that allows PHAs to track and contain COVID-19 infection based on users’ daily footprints without having direct access to them.

Keywords: Trusted Execution Environment, Private Set Intersection, Private Histogram Calculation

Relevant software: TBA


Optimization of lattice-based homomorphic encryption schemes

Lattice-based homomorphic encryption is gaining much industrial attraction due to its versatility and post-quantum security. It is, however, known for the notorious inefficiency due to the large cipher texts and complex operations. We study how different special parameter settings along with customized algorithms optimize the homomorphic encryption schemes in CPUs and compute-enabled RAM.

Keywords: Fully/Somewhat homomorphic encryption, hardware accelerator

Relevant software: Full-RNS B/FV scheme with Fermat/Mersenne numbers, Benchmarking Microsoft SEAL for comparing against compute-enabled RAM


Secure and accountable management of big data

More and more data are generated and collected nowadays, but we do not have a way to monitor and control the management of those data. We study how to let individuals hold ultimate controls over their own personal data being collected everyday and everywhere.

Keywords: Accountability, data provisioning, secure provenance

Relevant software: Secure fuzzy deduplication on images, ProvNet: Tracking data provenance with blockNet


Privacy-preserving distributed deep learning

Deep learning technologies have given birth to numerous innovative applications in our life, and it is expanding to individuals’ devices. We believe now it is a proper time to consider the user privacy implications behind this breakthrough technology. In this project, we study how to enable various deep learning technologies without breaching individual privacy in distributed/decentralized environments.

Keywords: Privacy-preserving computation, applied cryptography, secure multi-party computation, Trusted execution environment, secure aggregation

Relevant software: One-round secure multiparty computation with TPMModified HEtest framework for testing and comparing SEAL/HElib, Cryptonite: ECC-based secure aggregation, SGXNN: Neural network training with SGX & GPU


On scalability and maintenance cost of blockchain

Blockchain has various desirable security properties (e.g., tamper-proofness, decentralization), however it has several shortcomings as well. Our goal is to make blockchain more scalable and sustainable.

Keywords: Blockchain, efficiency, scalability

Relevant software: Blockchain with proof of deep learning




Collaboration

We have been actively collaborating with the following groups/labs.