SeDA: Secure and Efficient DNN Accelerators with Hardware/Software Synergy

TL;DR

SeDA is a novel DNN accelerator security framework that combines bandwidth-aware encryption, optimized tiling, and multi-level integrity verification to enhance security while reducing hardware and memory overheads.

Contribution

SeDA introduces a comprehensive security approach for DNN accelerators that significantly reduces performance overheads and hardware resource usage.

Findings

Reduces performance overhead by over 12%

Ensures robust scalability in secure DNN processing

Minimizes memory access overheads through multi-level verification

Abstract

Ensuring the confidentiality and integrity of DNN accelerators is paramount across various scenarios spanning autonomous driving, healthcare, and finance. However, current security approaches typically require extensive hardware resources, and incur significant off-chip memory access overheads. This paper introduces SeDA, which utilizes 1) a bandwidth-aware encryption mechanism to improve hardware resource efficiency, 2) optimal block granularity through intra-layer and inter-layer tiling patterns, and 3) a multi-level integrity verification mechanism that minimizes, or even eliminates, memory access overheads. Experimental results show that SeDA decreases performance overhead by over 12% for both server and edge neural processing units (NPUs), while ensuring robust scalability.