SGX-Stream: a secure stream analytics framework In SGX-enabled edge cloud

Kassem Bagher, Shangqi Lai

Research output: Contribution to journalArticleResearchpeer-review

1 Citation (Scopus)

Abstract

This paper introduces SGX-Stream, a secure and efficient data analytics framework for data streams using Intel SGX. SGX-Stream employs sketch algorithms in a cloud–edge architecture. To ensure performance and security, SGX-Stream preprocesses the data at the edge of the network to generate sketches and send them to the cloud for further processing inside the SGX enclave. To prioritise urgent tasks, SGX-Stream develops a hybrid task-aware scheduler tailored for SGX to manage task execution securely and practically. SGX-Stream is implemented as a full-fledged framework within the enclave with a small TCB size of 6 kLoC. With an extensible interface, SGX-Stream facilitates the development of various applications, such as adversarial attack detection over data stream. Under different workloads, SGX-Stream can bring 14× speedup on urgent tasks with less than 800 KB of scheduling memory consumption. We also demonstrate SGX-Stream's practicality with three real-world applications.

Original languageEnglish
Article number103403
Number of pages11
JournalJournal of Information Security and Applications
Volume72
DOIs
Publication statusPublished - Feb 2023

Cite this