Uncertainty-Aware Hardware Trojan Detection Using Multimodal Deep Learning

Rahul Vishwakarma,Amin Rezaei

2024-01-23

Abstract:The risk of hardware Trojans being inserted at various stages of chip production has increased in a zero-trust fabless era. To counter this, various machine learning solutions have been developed for the detection of hardware Trojans. While most of the focus has been on either a statistical or deep learning approach, the limited number of Trojan-infected benchmarks affects the detection accuracy and restricts the possibility of detecting zero-day Trojans. To close the gap, we first employ generative adversarial networks to amplify our data in two alternative representation modalities, a graph and a tabular, ensuring that the dataset is distributed in a representative manner. Further, we propose a multimodal deep learning approach to detect hardware Trojans and evaluate the results from both early fusion and late fusion strategies. We also estimate the uncertainty quantification metrics of each prediction for risk-aware decision-making. The outcomes not only confirms the efficacy of our proposed hardware Trojan detection method but also opens a new door for future studies employing multimodality and uncertainty quantification to address other hardware security challenges.

Cryptography and Security,Artificial Intelligence,Machine Learning

What problem does this paper attempt to address?

This paper attempts to solve several key problems in Hardware Trojan (HT) detection, especially in the zero - trust fabless era, where the risk of Hardware Trojans being inserted at various stages of chip production has increased. Specifically, the paper mainly focuses on the following points:

1. **The problem of insufficient data sets**: Existing Hardware Trojan detection methods rely on a limited number of known Trojan samples, which limits the detection accuracy and the ability to detect unknown Trojans (i.e., zero - day Trojans).

2. **Utilization of multi - modal data**: Most existing methods use only a single type of feature representation (such as statistical or deep - learning methods) and ignore the potential complementarity between different data representations. Therefore, these methods may not be able to make full use of all available information.

3. **Uncertainty quantification**: In practical applications, especially in safety - critical areas, the evaluation of the uncertainty of model predictions is crucial for making risk - aware decisions. However, existing Hardware Trojan detection methods usually lack effective quantification of uncertainty.

To solve these problems, the authors propose a Hardware Trojan detection framework named NOODLE based on multi - modal deep learning and uncertainty quantification. Specific measures include:

- Using Generative Adversarial Networks (GANs) to expand the data set to ensure the representativeness of the data distribution.
- Adopting two data representation forms, graphs and tables, and fusing them through a multi - modal deep - learning model to improve the detection accuracy.
- Introducing uncertainty quantification techniques, such as p - value aggregation and Conformal Prediction, to enhance the interpretability and reliability of the model.

Through these methods, the paper not only verifies the effectiveness of the proposed Hardware Trojan detection technology but also provides new directions for future research, especially on how to use multi - modal and uncertainty quantification to deal with other hardware security problems.