The original paper is in English. Non-English content has been machine-translated and may contain typographical errors or mistranslations. ex. Some numerals are expressed as "XNUMX".
Copyrights notice
The original paper is in English. Non-English content has been machine-translated and may contain typographical errors or mistranslations. Copyrights notice
Les circuits intégrés (CI) numériques modernes sont souvent conçus et fabriqués par des tiers et par des outils, ce qui peut rendre la conception/fabrication des CI vulnérables aux modifications malveillantes. Les circuits malveillants sont généralement appelés chevaux de Troie matériels (HT) et sont considérés comme constituant un problème de sécurité sérieux. Dans cet article, nous proposons une méthode de détection et de classification HT basée sur des tests logiques utilisant l'apprentissage en régime permanent. Nous observons d’abord que les HT sont masqués lors de l’application de modèles de test aléatoires sur un court laps de temps, mais que la plupart d’entre eux peuvent être activés lors d’un fonctionnement de circuit aléatoire à très long terme. Il est donc très naturel que nous apprenions états de transition de signal stables de chaque réseau de chevaux de Troie suspect dans une netlist en effectuant une simulation aléatoire à court terme. Après cela, nous simulons ou émulons la netlist sur une très longue période en donnant des modèles de test aléatoires et obtenons un ensemble d'états de transition de signal. En découvrant la corrélation entre eux, notre méthode détecte les HT et découvre son comportement. Parfois, les HT n’affectent pas les sorties principales mais diffusent simplement des informations via des canaux secondaires. Notre méthode peut être appliquée avec succès à ces types de HT. Les résultats expérimentaux démontrent que notre méthode peut identifier avec succès tous les réseaux de chevaux de Troie réels comme étant des réseaux de chevaux de Troie et tous les réseaux normaux comme étant des réseaux normaux, alors que d'autres méthodes de détection HT existantes ne peuvent pas détecter certains d'entre eux. De plus, notre méthode peut détecter avec succès les HT même s’ils ne sont pas réellement activés lors d’une simulation aléatoire à long terme. Notre méthode devine également correctement le comportement HT en utilisant l'apprentissage de la transition du signal.
Masaru OYA
Waseda University
Masao YANAGISAWA
Waseda University
Nozomu TOGAWA
Waseda University
The copyright of the original papers published on this site belongs to IEICE. Unauthorized use of the original or translated papers is prohibited. See IEICE Provisions on Copyright for details.
Copier
Masaru OYA, Masao YANAGISAWA, Nozomu TOGAWA, "Hardware Trojan Detection and Classification Based on Logic Testing Utilizing Steady State Learning" in IEICE TRANSACTIONS on Fundamentals,
vol. E101-A, no. 12, pp. 2308-2319, December 2018, doi: 10.1587/transfun.E101.A.2308.
Abstract: Modern digital integrated circuits (ICs) are often designed and fabricated by third parties and tools, which can make IC design/fabrication vulnerable to malicious modifications. The malicious circuits are generally referred to as hardware Trojans (HTs) and they are considered to be a serious security concern. In this paper, we propose a logic-testing based HT detection and classification method utilizing steady state learning. We first observe that HTs are hidden while applying random test patterns in a short time but most of them can be activated in a very long-term random circuit operation. Hence it is very natural that we learn steady signal-transition states of every suspicious Trojan net in a netlist by performing short-term random simulation. After that, we simulate or emulate the netlist in a very long time by giving random test patterns and obtain a set of signal-transition states. By discovering correlation between them, our method detects HTs and finds out its behavior. HTs sometimes do not affect primary outputs but just leak information over side channels. Our method can be successfully applied to those types of HTs. Experimental results demonstrate that our method can successfully identify all the real Trojan nets to be Trojan nets and all the normal nets to be normal nets, while other existing logic-testing HT detection methods cannot detect some of them. Moreover, our method can successfully detect HTs even if they are not really activated during long-term random simulation. Our method also correctly guesses the HT behavior utilizing signal transition learning.
URL: https://global.ieice.org/en_transactions/fundamentals/10.1587/transfun.E101.A.2308/_p
Copier
@ARTICLE{e101-a_12_2308,
author={Masaru OYA, Masao YANAGISAWA, Nozomu TOGAWA, },
journal={IEICE TRANSACTIONS on Fundamentals},
title={Hardware Trojan Detection and Classification Based on Logic Testing Utilizing Steady State Learning},
year={2018},
volume={E101-A},
number={12},
pages={2308-2319},
abstract={Modern digital integrated circuits (ICs) are often designed and fabricated by third parties and tools, which can make IC design/fabrication vulnerable to malicious modifications. The malicious circuits are generally referred to as hardware Trojans (HTs) and they are considered to be a serious security concern. In this paper, we propose a logic-testing based HT detection and classification method utilizing steady state learning. We first observe that HTs are hidden while applying random test patterns in a short time but most of them can be activated in a very long-term random circuit operation. Hence it is very natural that we learn steady signal-transition states of every suspicious Trojan net in a netlist by performing short-term random simulation. After that, we simulate or emulate the netlist in a very long time by giving random test patterns and obtain a set of signal-transition states. By discovering correlation between them, our method detects HTs and finds out its behavior. HTs sometimes do not affect primary outputs but just leak information over side channels. Our method can be successfully applied to those types of HTs. Experimental results demonstrate that our method can successfully identify all the real Trojan nets to be Trojan nets and all the normal nets to be normal nets, while other existing logic-testing HT detection methods cannot detect some of them. Moreover, our method can successfully detect HTs even if they are not really activated during long-term random simulation. Our method also correctly guesses the HT behavior utilizing signal transition learning.},
keywords={},
doi={10.1587/transfun.E101.A.2308},
ISSN={1745-1337},
month={December},}
Copier
TY - JOUR
TI - Hardware Trojan Detection and Classification Based on Logic Testing Utilizing Steady State Learning
T2 - IEICE TRANSACTIONS on Fundamentals
SP - 2308
EP - 2319
AU - Masaru OYA
AU - Masao YANAGISAWA
AU - Nozomu TOGAWA
PY - 2018
DO - 10.1587/transfun.E101.A.2308
JO - IEICE TRANSACTIONS on Fundamentals
SN - 1745-1337
VL - E101-A
IS - 12
JA - IEICE TRANSACTIONS on Fundamentals
Y1 - December 2018
AB - Modern digital integrated circuits (ICs) are often designed and fabricated by third parties and tools, which can make IC design/fabrication vulnerable to malicious modifications. The malicious circuits are generally referred to as hardware Trojans (HTs) and they are considered to be a serious security concern. In this paper, we propose a logic-testing based HT detection and classification method utilizing steady state learning. We first observe that HTs are hidden while applying random test patterns in a short time but most of them can be activated in a very long-term random circuit operation. Hence it is very natural that we learn steady signal-transition states of every suspicious Trojan net in a netlist by performing short-term random simulation. After that, we simulate or emulate the netlist in a very long time by giving random test patterns and obtain a set of signal-transition states. By discovering correlation between them, our method detects HTs and finds out its behavior. HTs sometimes do not affect primary outputs but just leak information over side channels. Our method can be successfully applied to those types of HTs. Experimental results demonstrate that our method can successfully identify all the real Trojan nets to be Trojan nets and all the normal nets to be normal nets, while other existing logic-testing HT detection methods cannot detect some of them. Moreover, our method can successfully detect HTs even if they are not really activated during long-term random simulation. Our method also correctly guesses the HT behavior utilizing signal transition learning.
ER -