日時: 9月22日(木)1限(9:20-10:50)

会場: L1

司会: KAN Yirong
西山 輝 D, 中間発表 情報セキュリティ工学 林 優一, 岡田 実, 藤川 和利, 藤本 大介, KIM Youngwoo
title: Evaluation of the Impacts of Intentional Electromagnetic Interference against Cryptographic Circuits and its Countermeasures
abstract:Intentional electromagnetic interference (IEMI) against cryptographic circuits has been reported as a fault injection attack, which induces temporal calculation error (fault) only at specific timing in the cryptographic operations to extract secret key. Among the conventional fault injection methods, a method that does not require physical access and/or invasion of the cryptographic devices has been reported by injecting EM waves through the power cable of the devices. However, since this method assumes the attacker doesn't have the timing information at which the cryptographic operation executes from a distance, it is difficult to generate faulty outputs only at a specific timing. In this case, a large number of cryptographic operations must be executed to obtain the faulty outputs necessary for key extraction, limiting the scope of threat application. In contrast, if it is possible to generate faults necessary for key extraction with a small number of cryptographic operations even under the condition that the timing information is unknown, the scope of threats would be expanded. This study evaluates the impact of IEMI-based fault injection attack using EM waves from a distance, and builds a design methodologies for cryptographic circuits resistance to IEMI-based fault injection attack based on the evaluation. language of the presentation: Japanese
尾崎 慧一 M, 2回目発表 情報セキュリティ工学 林 優一, 岡田 実, 藤川 和利, 藤本 大介, KIM Youngwoo
​ title: A Study on Randomness Evaluation of RO-based TRNG Using Amplitude Probability Distribution ​
​ abstract: Randomness evaluation of true random number generators (TRNGs) based on ring oscillators (ROs) requires an analysis of output sequences from TRNGs. However, direct evaluation of the sequences is difficult because the sequences are an intermediate value when the TRNGs are implemented in a cryptographic module. This research proposes a method to predict randomness degradation from radiated electromagnetic waves of TRNGs. Specifically, an amplitude probability distribution (APD) used for interference effects evaluation of electromagnetic interference against RO-based TRNGs is measured. In addition, it is verified that the features extracted from APD waveforms depend on entropy.
​ language of the presentation: Japanese
富島 み登り M, 2回目発表 情報セキュリティ工学 林 優一, 安本 慶一, 藤川 和利, 藤本 大介, KIM Youngwoo
title: A Determination Method of the Irradiation Timing of an Attack Pulse for Evaluation of Distance Spoofing Attack Against direct ToF LiDAR
abstract: This research proposes a determination method of the irradiation timing of an attack pulse and verifies that the same system is capable of evaluating the attack against different models of the LiDAR. Previous researches use “one” photodiode in order to observe the timing of a LiDAR measuring. In this study, “two” photodiodes are used, one at the observation location and the other at the attack location. The delay time required for the attack is calculated based on the time difference between two locations. The proposed method is applied to other models of LiDAR and showed that the attack is valid regardless of its model.
language of the presentation: Japanese
発表題目: Direct ToF LiDARに対する距離偽装攻撃耐性評価における攻撃光照射タイミング決定法に関する研究
発表概要: 本研究では、攻撃光を照射するタイミングの決定法を提案し、異なるLiDARに対しても同一の評価系で攻撃評価を実現する。 具体的には、LiDARの計測タイミングを観測するために使用されてきたフォトダイオードの数を従来の1つから2つに増やすことで、観測地点および攻撃地点の2地点間の時間差から攻撃に必要な遅延時間を算出する。 複数台のLiDARに本手法を適用し、LiDARによらず攻撃が成立することを示す。
芳賀 陸雄 M, 2回目発表 情報セキュリティ工学 林 優一, 中島 康彦, 藤川 和利, 藤本 大介, KIM Youngwoo
title: A Study on Fast and Small-area CSIDH Implementation using RNS-based Parallel Architecture.
In recent years, research is being conducted on a post quantum cryptography (PQC), which is secure against quantum computers."Isogeny" based commutative supersingular isogeny diffie-hellman(CSIDH) is proposed as one of the PQC candidates.The CSIDH has attracted attention because its key size is much smaller than other candidates, and thus it can be implemented in resource-restricted devices. However, the CSIDH has a slower execution time than other candidates due to large computational complexity of algorithm and large finite field values . This study aims to propose fast and small-area CSIDH implementation using the residue number system which is suitable for parallel computation.
language of the presentation: Japanese
発表題目: RNSを用いた並列アーキテクチャによるCSIDHの高速かつ小面積な実装に関する研究
発表概要: 現在,量子計算機に対しても安全な耐量子計算機暗号( PQC )についての研究が行われている.PQCの候補の1つに「超特異楕円曲線間の同種写像問題」を安全性の根拠とするCommutative Supersingular Isogeny Diffie-Hellman ( CSIDH )という鍵共有方式が提案されている.CSIDHは他のPQC候補に比べて鍵のデータサイズが非常に小さいことから,回路面積や消費電力などの資源制約のあるデバイスへPQCを搭載できる可能性があることで注目されている.しかし,CSIDHはアルゴリズムの計算量が大きいことや扱う有限体の値が大きいことから他のPQC候補よりも実行時間が遅いことが課題となっている.本研究では,並列アーキテクチャに適した整数表現であるResidue Number Systemを用いたハードウェア実装によるCSIDHの高速かつ小面積な実装法を提案することを目指す.

会場: L2

司会: 福嶋 誠
伊野 淳也 M, 2回目発表 生体医用画像 佐藤 嘉伸, 加藤 博一, 大竹 義人, Soufi Mazen
title: Evaluation of Patient-specific musculoskeletal simulation using MRI images in multiple pose
abstract: We aim at developing a patient-specific musculoskeletal simulation approach for predicting musculoskeletal behavior of musculoskeletal structures after hip surgery, disease progression, and rehabilitation. In this research, we performed patient-specific musculoskeletal modeling of using elastic volume simulator (VIPER) based on geometric representation of the muscle shape and muscle fiber orientation derived from volumetric medical images. We focused on the hip joint (the pelvis, part of the femur, and the gluteus medius) as a target for quantitative evaluation using standing MRI images of multiple patients in three poses: neutral, abduction, and rotation. We constructed patient-specific musculoskeletal models using the MRI images in the neutral pose, and predicted geometric representation of the muscle shape and muscle fiber orientation in abduction and rotation poses. We report the results of quantitative evaluation by comparing the prediction results with MRI images of each poses.
language of the presentation: Japanese
下元 悠我 M, 2回目発表 生体医用画像 佐藤 嘉伸, 金谷 重彦, 大竹 義人, Soufi Mazen
title: Construction of a statistical model for predicting aging changes in the pelvis and spine using a large CT image database.
abstract: A whole-body skeletal model that is faithful to the human body is important for skeletal structural and movement analysis of the human body. Currently, a statistical shape model (SSM), which statistically analyzes the shape data of organs of multiple individuals, is used as one of the methods to represent the skeleton. However, SSMs have been constructed only for a part of the skeletal shape in previous studies, but not for the whole body skeleton. Therefore, we aim to construct a whole-body skeletal SSM using the J-MID database collected by the Japan Radiological Society and a large CT database of more than 40,000 cases collected at our collaborating institutions. As a first step, we will construct SSMs using polygonal models and landmarks of the torso using large-scale data of the torso. We analyze the aging change using the constructed SSM and predict the skeletal shape from only the pelvic region using the SSM of landmarks.
language of the presentation: Japanese
発表題目: 大規模CT画像データベースを用いた骨盤・脊椎の加齢変化予測統計モデルの構築
発表概要: 人体に忠実な全身骨格モデルは、人体の骨格構造解析や動作解析において重要である。現在、骨格の表現方法の一つとして、複数人の臓器の形状データを統計的に解析した統計形状モデル(SSM)が用いられている。しかし、従来研究では一部の骨格形状のみのSSMを構築しており、全身骨格の構築は行われていない。その為、我々は日本医学放射線学会が収集しているJ-MIDデータベースと、共同研究施設で収集する4万症例以上の大規模なCTデータベースを用いた全身骨格のSSM構築を目指している。本研究ではその第一歩として、体幹部の大規模データを用いてポリゴンモデルで表現したSSM構築と、ランドマークによって表現したSSM構築を行う。そして、構築したSSMで加齢変化の解析を行うと共に、ランドマークのSSMを用いて骨盤周辺部のみから骨格形状の予測を行ったので、その結果を示す。
箱谷 知輝 M, 2回目発表 生体医用画像 佐藤 嘉伸, 池田 和司, 大竹 義人, Soufi Mazen
title: Virtual Standing Position Transformation from Supine CT using Whole-body Upright CT Scanner
abstract: The modeling of human motion relies on the accurate representation of the deformations in the musculoskeletal geometry that can be captured by medical images. However, the relative positions and shapes of the musculoskeletal structures change depending on the body positioning during the image acquisition. Volumetric medical images, e.g. CT images, are usually acquired in the supine positioning, thus they may not reflect the actual deformation occurring under the standing position. Meanwhile, due to the scarcity of standing CT imaging systems, the creation of models based on the musculoskeletal geometry in the standing position is non-trivial. Therefore, the aim of this research is to create a statistical deformation model that predicts the deformations in the standing position from images scanned in the supine position. The 3D shape models of the target musculoskeletal structures were automatically segmented from supine and standing CT images. The muscle deformations from supine to standing positions were modeled using a partial least square regression model. In this study, the proposed model was developed and validated on a database of 32-pairs of supine and standing CT images of 32 healthy volunteers.
language of the presentation: Japanese