On design vulnerability analysis and trust benchmarks development

TLDR

Hardware security and trust have grown rapidly, yet Trojan detection research suffers from a lack of standard benchmarks and measurements, leading to inconsistent outcomes and unclear assessments of technique strengths and weaknesses. The authors propose a vulnerability analysis flow that identifies hard‑to‑detect circuit regions likely to host Trojans and introduces a detectability metric to quantify Trojan activation and impact. These methods produced a large public benchmark suite that facilitates fair comparison of Trojan detection techniques and advances the field.

Abstract

The areas of hardware security and trust have experienced major growth over the past several years. However, research in Trojan detection and prevention lacks standard benchmarks and measurements, resulting in inconsistent research outcomes, and ambiguity in analyzing strengths and weaknesses in the techniques developed by different research teams and their advancements to the state-of-the-art. We have developed innovative methodologies that, for the first time, more effectively address the problem. We have developed a vulnerability analysis flow. The flow determines hard-to-detect areas in a circuit that would most probably be used for Trojan implementation to ensure a Trojan goes undetected during production test and extensive functional test analysis. Furthermore, we introduce the Trojan detectability metric to quantify Trojan activation and effect. This metric offers a fair comparison for analyzing weaknesses and strengths of Trojan detection techniques. Using these methodologies, we have developed a large number of trust benchmarks that are available for use by the public, as well as researchers and practitioners in the field.