| BORHAN ROWNAK | M, 2回目発表 | サイバーレジリエンス構成学 | 門林 雄基 | 和田 隆広 | 妙中 雄三 | |
|
title: Optimized CRYSTALS-Dilithium: Enhancing Performance and Security for IoT in the Quantum Age.
abstract: Quantum computing is on the cusp of practical deployment, bringing unparalleled computational power and problem-solving capabilities across various fields. However, this progress presents a critical challenge to the security of existing cryptographic systems. To address this threat, efforts have been directed toward establishing quantum-resistant cryptographic standards. The National Institute of Standards and Technology (NIST) has selected CRYSTALS-Dilithium as a prominent lattice- based digital signature algorithm through a rigorous standard- ization process. While advancements in computational efficiency, particularly in polynomial multiplication, have been achieved, research on optimizing the hashing process and security concerns during CRYSTALS-Dilithium’s digital signature is limited. To overcome this limitation, this study introduces a novel approach that replaces the SHAKE function in CRYSTALS-Dilithium with a combination of IOSHA and AES, tailored for Internet of Things (IoT) applications. By leveraging dynamic parameters and pseudorandom number-based IOSHA, this method significantly improves computational speed on Advanced RISC Machine (ARM) Cortex-M7 processors while enhancing security compared to SHAKE-256. The proposed integration not only accelerates processing but also bolsters resistance against MLWE, MSIS, and SelfTargetMSIS challenges, all without necessitating hardware upgrades. This research highlights a significant step forward in developing efficient and secure quantum-resistant cryptographic protocols. language of the presentation: English | ||||||
| VALERIE MEGAN | M, 2回目発表 | サイバーレジリエンス構成学 | 門林 雄基 | 笠原 正治 | 妙中 雄三 | |
|
title: Detecting Malicious AI-Generated Personas Through Hybrid Systems Using Rule-Based Logic and Machine Learning-Assisted Models
abstract: The rapid evolution of AI technologies has led to the emergence of AI-generated personas--digital entities designed to mimic human behavior with increasing sophistication. While these personas are primarily used for marketing and entertainment, they pose significant risks, including manipulation, misinformation, and identity deception. Current detection approaches focus predominantly on generated content rather than the personas as holistic entities. This research bridges the gap by proposing a hybrid detection system integrating rule-based logic with machine learning models. Our methodology leverages a novel taxonomy of harmfulness to analyze personas' digital footprints. Preliminary results using our dataset of benign, subtly malicious, and explicitly malicious personas demonstrate the feasibility of detecting harmful personas, even with subtle intent. This study highlights the importance of developing lightweight, robust tools to safeguard online communities against the rising threats posed by AI-generated personas. language of the presentation: English | ||||||
| 名越 遼 | M, 1回目発表 | 情報セキュリティ工学 | 林 優一 | 岡田 実 | 向川 康博 | 藤本 大介 |
|
title: Study on Information Leakage Induced by Multiple Frequency Electromagnetic Waves Irradiation via Nonlinear Element Frequency Conversion
abstract: Information processing within digital systems is manifested through electrical signals, whose temporal fluctuations generate secondary electromagnetic radiation. These unintended emissions are susceptible to interception and analysis, constituting a significant security vulnerability. While devices exhibiting low emission intensities have traditionally been considered resistant to such vulnerabilities due to detection limitations, emerging research has identified novel security concerns. These concerns stem from enhanced emission intensities induced by intentionally applied external electromagnetic radiation, even in devices previously deemed secure against conventional threats. A fundamental challenge in this approach arises from spectral interference between incident and reflected waves, which compromises signal recovery fidelity. This investigation explores methodologies to mitigate interference effects and broaden the scope of potential attack vectors. We propose a novel technique leveraging frequency conversion phenomena in nonlinear circuit elements to attenuate interference between incident and reflected waves. Furthermore, experimental validation using operational information processing systems demonstrates the practical viability of this proposal method. language of the presentation: Japanese | ||||||
| 松本 匠平 | M, 1回目発表 | 情報セキュリティ工学 | 林 優一 | 岡田 実 | 向川 康博 | 藤本 大介 |
|
title: Investigation of Methods for Extracting Video Information from Radiated Electromagnetic Waves Using Nearby Metallic Objects
abstract: With the spread of digital devices, displays are increasingly being used to handle sensitive information, such as authentication and account details. Electromagnetic waves generated during display operation may leak and expose sensitive information, thereby posing significant security risks. Attackers can receive these waves and subsequently reconstruct display images, potentially compromising confidential data. In response to this threat, a countermeasure known as zoning has been implemented and extensively considered. Zoning effectively attenuates the leaked electromagnetic waves to levels below background noise on the transmission path, thus rendering the attack difficult. However, portable attack configurations enable attackers to measure waves from closer proximities, thereby potentially nullifying zoning countermeasures. Therefore, this study investigates methodologies for enhancing the portability of antennas, which constitutes one of the key factors in constructing a portable attack configuration. language of the presentation: Japanese | ||||||