Title: Implementation and Analysis of Fully Homomorphic Encryption in Wearable Devices

Year of Publication: Dec - 2018
Page Numbers: 1-14
Authors: Amonrat Prasitsupparote, Yohei Watanabe, Junji Shikata
Conference Name: The Fourth International Conference on Information Security and Digital Forensics (ISDF2018)
- Greece

Abstract:


Currently, wearable devices, which are known as one of the Internet of things (IoT) devices, have been widely used for healthcare systems. Most of the healthcare systems store users' healthcare data, which is encrypted by ordinary symmetric-key encryption and/or public-key encryption schemes, in a (cloud) server. However, the encrypted data must be decrypted for data analysis, and it means that sensitive information is leaked to the server. One promising solution is to use fully homomorphic encryption (FHE), which enables ones to perform any computation among encrypted data while keeping it encrypted. Although FHE generally requires high computational and communication costs in the theoretical sense, several researchers have implemented FHE schemes to measure their practical efficiency. In this paper, we consider a privacy preserving protocol for healthcare systems employing wearable devices, and implement this protocol over Raspberry Pi, which is a well-known single-board computer, to measure the actual efficiency of FHE over wearable devices. Specifically, we implemented the protocol by using two FHE libraries, \textit{HElib} and \textit{SEAL}, on Raspberry Pi and network simulator to measure both computational and communication costs in wireless body area network (WBAN). In terms of the communication overhead, our results show that the protocol with SEAL is better than that with HElib. In particular, the protocol with SEAL has almost the same communication costs as the \textit{trivial} protocol, which is the same protocol without encryption. On the other hand, HElib is better than SEAL in terms of the running time, while SEAL can perform better homomorphic operations than HElib for the same plaintext-size. Therefore, HElib is suitable for applications which require small time complexity, and SEAL is suitable for applications which require many homomorphic operations.