Constrained PRFs (CPRFs) have many applications. Until recently, we only knew how to construct CPRFs for simple constraint predicates under standard assumptions. We examine the case of inner product predicates (which give rise to a number of other useful predicates such as predicates described by constant-degree polynomials). We show that it is possible to construct constraint-hiding CPRFs for inner product predicates (1) unconditionally in the random oracle model, (2) under DDH, and (3) from the minimal assumption that one-way functions exist. Previously, CPRFs for inner product constraints were only known from the DCR and LWE assumptions, or from non-standard assumptions. Our constructions are also the first to be concretely practical.

Paper · Slides · Code Function secret sharing (FSS) has seen many applications in privacy-preserving systems. An important feature in these systems is the ability to enforce access control privately. We formalize this notion and provide several constructions for access control in FSS. We evaluate several applications of access control for FSS, including PIR with access control and faster anonymous communication.

Paper · Slides · Code

Talk given at: IRIF laboratory at the CNRS and University of Paris Cité, IEEE S&P. Mix-nets offer great scalability properties and can be used to instantiate anonymous broadcast (and communication) efficiently. We design and build Trellis, a new mix-net based anonymous broadcast system that offers: (1) concrete efficiency, (2) network robustness in the face of malicious servers, (3) scalability with added servers, and (4) security even when a majority of mix servers are malicious.

Paper · Slides · Code Tor can be significantly slower compared to regular browsing largely due to the way in which Tor routes traffic over the internet. Tor requires routing packets through multiple relays forming a "circuit." Traffic congestions between relays on a Tor circuit can lead to delays, increasing latency, and hindering user experience. ShorTor is an overlay for the Tor network which can help find shorter paths between relays using a trick deployed by major CDNs. ShorTor reduces latency between relays, is incrementally deployable, and minimally impacts the security of Tor.

Paper · Slides · Code

• Private Access Control for Function Secret Sharing
  Sacha Servan-Schreiber, Simon Beyzerov, Eli Yablon, and Hyojae Park
  IEEE Symposium on Security and Privacy (S&P) 2023.
  
• Trellis: Robust and Scalable Metadata-private Anonymous Broadcast
  Simon Langowski, Sacha Servan-Schreiber, and Srini Devadas
  Network and Distributed System Security Symposium (NDSS) 2023.
  
• Designing Hardware for Cryptography and Cryptography for Hardware
  Srini Devadas, Simon Langowski, Nikola Samardzic, Sacha Servan-Schreiber, and Daniel Sanchez
  ACM SIGSAC Conference on Computer and Communications Security (CCS) 2022. (Invited talk paper).
  
• ShorTor: Improving Tor Network Latency via Multi-hop Overlay Routing
  Kyle Hogan, Sacha Servan-Schreiber, Zachary Newman, Ben Weintraub, Cristina Nita-Rotaru, and Srini Devadas
  IEEE Symposium on Security and Privacy (S&P) 2022.
  
• Private Approximate Nearest Neighbor Search with Sublinear Communication
  Sacha Servan-Schreiber, Simon Langowski, and Srini Devadas
  IEEE Symposium on Security and Privacy (S&P) 2022.
  
• Spectrum: High-Bandwidth Anonymous Broadcast
  Zachary Newman, Sacha Servan-Schreiber, and Srini Devadas
  USENIX Symposium on Networked Systems Design and Implementation (NSDI) 2022.
  
• SchengenDB: A Data Protection Database Proposal
  Tim Kraska, Michael Stonebraker, Michael Brodie, Sacha Servan-Schreiber, Daniel Weitzner
  Heterogeneous Data Management, Polystores, and Analytics for Healthcare,
  VLDB Workshops, Poly 2020 and DMAH 2020.
• ProSecCo: Progressive Sequence Mining with Convergence Guarantees
  Sacha Servan-Schreiber, Matteo Riondato, and Emanuel Zgraggen
  Knowledge and Information Systems (KAIS) 2019. (Journal article).
  
• ProSecCo: Progressive Sequence Mining with Convergence Guarantees
  Sacha Servan-Schreiber, Matteo Riondato, and Emanuel Zgraggen
  IEEE International Conference on Data Mining (ICDM) 2018.
  Best Student Paper runner-up Award
• Towards Quantifying Uncertainty in Data Analysis & Exploration
  Yeounoh Chung, Sacha Servan-Schreiber, Emanuel Zgraggen, Tim Kraska
  IEEE Data Engineering Bulletin 2018.