Hanjun Kim  

Associate Professor
School of Electrical and Electronic Engineering, Yonsei University

Ph.D. 2013, Department of Computer Science, Princeton University

Office: Engineering Hall #3-C415
Phone: +82-2-2123-2770
Email: first_name at yonsei.ac.kr
 
 
[Home]   [Curriculum Vitae]   [Publications]   [CoreLab]   [Korean]  

Refereed International Conference Poster

Privacy Authority-Aware Compiler for Homomorphic Encryption on Edge-Cloud [abstract]
Dongkwan Kim, Yongwoo Lee, Seonyoung Cheon, Heelim Choi, Jaeho Lee, Dongyoon Lee, and Hanjun Kim
32nd USENIX Security Symposium - (Poster) (USENIX Security), August 2023.

For privacy preserving cloud computing, fully homomorphic encryption (FHE) is a promising solution by allowing cloud servers to compute on encrypted data without decryption. However, its programming burden on ciphertext management and huge latency of FHE operations limit its applicability. Recently proposed FHE compilers automate the ciphertext management, but they transform all the data into ciphertexts without filtering private data, thus increasing unnecessary FHE operations. Moreover, the compilers transform the entire program into an FHE program without partitioning, while computation on edges without encryption for private data can reduce the overall latency. This work introduces a new data type system, called privacy-set (PSet), that annotates devices with privacy authority for each private data. Moreover, this work proposes a new privacy authority-aware compiler, called PSet compiler that automatically transforms a PSet-annotated plain program into FHE-enabled edge-cloud cooperative program with operation latency- and communication-aware partitioning. This work evaluates the PSet compiler with six machine learning and deep learning applications and shows that the PSet compiler achieves 3.77 times faster than the existing FHE compiler without the partitioning support.