A Review and Comparison of AI Enhanced Side Channel Analysis

TL;DR

The paper analyzes how deep learning enhances side-channel analysis, focusing on profiling attacks that exploit leakage from power and EM channels to recover cryptographic keys. It evaluates state-of-the-art DL SCA techniques using the ASCAD dataset, comparing intra-device and cross-device approaches, and introduces the Key Recovery Difficulty metric to capture both attack success and data requirements. The survey highlights emerging directions, including non-profiled DL SCA, artificial trace generation, transferable models, and embeddings, and discusses implications for defenses such as hiding and masking. Overall, the work clarifies current capabilities and limitations of DL-based SCA, guiding future research toward more robust hardware security in the face of AI-powered attacks.

Abstract

Side Channel Analysis (SCA) presents a clear threat to privacy and security in modern computing systems. The vast majority of communications are secured through cryptographic algorithms. These algorithms are often provably-secure from a cryptographical perspective, but their implementation on real hardware introduces vulnerabilities. Adversaries can exploit these vulnerabilities to conduct SCA and recover confidential information, such as secret keys or internal states. The threat of SCA has greatly increased as machine learning, and in particular deep learning, enhanced attacks become more common. In this work, we will examine the latest state-of-the-art deep learning techniques for side channel analysis, the theory behind them, and how they are conducted. Our focus will be on profiling attacks using deep learning techniques, but we will also examine some new and emerging methodologies enhanced by deep learning techniques, such as non-profiled attacks, artificial trace generation, and others. Finally, different deep learning enhanced SCA schemes attempted against the ANSSI SCA Database (ASCAD) and their relative performance will be evaluated and compared. This will lead to new research directions to secure cryptographic implementations against the latest SCA attacks.

Figures (6)

• Figure 1: An example neuron in an artificial neural network.
• Figure 2: An example of a neural network with a hidden layer i.e. a deep learning network.
• Figure 3: A simple example of kernel convolution.
• Figure 4: The two stages of a Profiling Attack. In Step 1, the adversary trains a neural network (NN) using a device with know internal states or labels. In step two, the attack uses the NN to recover the internal states or labels of an uncontrolled device.
• Figure 5: Measurements to Disclose from the cited works which target ASCAD, lower is better. Each column is the name of the model provided by the work and its reference in our paper. The image above uses a linear scale, while the lower one uses a logarithmic scale, base 10.