| 渡辺 諒 | 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 | 鍛治 秀伍 |
|
title: A study on method for detecting aging of ICs using changes in backscatter from I/O circuit
abstract: The contamination of counterfeit semiconductors in electronic devices has become a serious issue. Particularly, recycled ICs which is removed from discarded devices are functionally equivalent to fresh ICs, making them difficult to detect through functional verification. In response to this issue, a method for detecting recycled ICs has been proposed that utilizes changes in electrical characteristics caused by backscattering generated in I/O circuits. However, the conventional method has focused on the amplitude of backscattering's harmonics to determine the applied frequency, which may not accurately capture the degradation effects in I/O circuits. Therefore, this study proposes a backscattering-based method for detecting aging focused on the degradation mechanisms in I/O circuits. This presentation shows an evaluation method for detecting degradation in PMOS and NMOS components within I/O circuits by analyzing differences in the backscattering's amplitudes. language of the presentation: Japanese | |||||||
| 大貫 和基 | M, 2回目発表 | 情報セキュリティ工学 | 林 優一 | 岡田 実 | 安本 慶一 | 藤本 大介 | |
|
title:A Study on Efficient Fault Analysis Using Side-Channel Information
abstract:Fault attacks on cryptographic modules consist of two phases: fault injection, which induces errors in encryption, and key analysis, which extracts the secret key from these calculation errors. While key analysis is effective when specific errors are induced during the injection phase, existing injection methods often produce various types of errors unsuitable for analysis. Consequently, attack occasions are limited. Our approach utilizes side-channel information, which reflects internal processes including faults, to discuss the possibility of expanding the threat range of fault attacks. As an example, we present an estimation method to determine which faults among those collected are suitable for analysis, using side-channel leakage, which indicates potential of reducing calculation cost. language of the presentation: Japanese 発表題目:暗号モジュールのサイドチャネル情報を用いた故障利用解析の効率化に関する研究 発表概要:暗号モジュールに対する故障利用攻撃は、暗号処理に計算誤りを引き起こす故障注入と、その際に得られる誤り出力から秘密鍵を取得する秘密鍵解析の二つの要素から構成される。秘密鍵解析では、故障注入時における特定の誤りの発生を仮定しているが、既存の故障注入法では、この仮定に合致しない誤りが多く発生する。そのため故障利用攻撃が成立する条件は限定的であった。これに対し本研究では、誤り発生に伴う内部処理の変化をサイドチャネル情報から取得することで、故障利用攻撃の効率が向上する可能性について議論をする。その一例として本発表では、サイドチャネル情報を用いることで、得られる誤り出力が秘密鍵解析に適用可能か否かを推定し、解析の際の計算量が削減される可能性を示す。 | |||||||
| 池内 陸 | M, 2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: A Study on the Evaluation of Electromagnetic Information Leakage from Equipment with Encrypted Transmission Signals.
abstract: There is a known threat of input/output information handled by electronic devices being leaked by unwanted electromagnetic radiation. In order to mitigate this threat, a method that makes it difficult for an attacker to restore information by using encryption of transmission signals at the upper layer is estimated. On the other hand, as the encryption is decrypted inside the equipment, unwanted electromagnetic radiation from the line transmitting the decrypted signal may become an attack target. Therefore, this study proposes a method to assess leakage from the lines transmitting the decrypted signals to the encrypted equipment. The presented method is based on an investigation of the recovery of screen information and the leakage paths of signals containing screen information for displays with a mixture of transmission lines where screen information encryption has been applied and unapplied. language of the presentation: Japanese 発表題目: 伝送信号を暗号化した機器からの電磁的情報漏えい評価に関する研究 発表概要: 電子機器で扱われる入出力情報が不要放射を通じて漏えいする脅威が知られている。本脅威の対策として、上位層における伝送信号の暗号化を利用することで、攻撃者の情報復元を困難化する手法が想定されている。一方で、暗号化は機器の内部で復号されるため、復号された信号を伝送する線路からの不要放射が攻撃対象になる可能性がある。そこで、本研究では、暗号化された機器に対して、復号された信号を伝送する線路からの漏えいについて評価する手法を提案する。本発表では、画面情報の暗号化が適用された伝送路と未適用の伝送路が混在するディスプレイを対象に、画面情報の復元と画面情報を含む信号の漏えい経路の調査を行った。 | |||||||
| 角野 匠 | M, 2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: A Study on Countermeasure Using Noise Addition in Electromagnetic Sound Information Leakage
abstract: The threat of leakage of I/O information of electronic devices through electromagnetic waves is known. To counter such threats, it is necessary to prevent leakage through electromagnetic waves without losing any information. However, no countermeasures have been considered for real-time information, such as audio data. Therefore, the purpose of this study is to develop a countermeasure method that makes it difficult to restore information through electromagnetic waves without disturbing the user's ease of hearing by using noise addition to the sound source as an information change method. In this presentation, we examined a countermeasure method where two audio signals with inverse-phase noise were added to the audio source and presented using stereo functionality, resulting in the cancellation of the noise for the listener. language of the presentation: Japanese 発表題目: 電磁的音情報漏えいにおけるノイズ付加を利用した対策手法に関する研究 発表概要: 電子機器の入出力情報が電磁波を通じて漏えいする脅威が知られている。このような脅威への対策としては、情報を欠落することなく、電磁波を通じた漏えいのみを防ぐ必要がある。しかし、音情報のようなリアルタイム情報においては対策が検討されていない。そこで、本研究では、情報の変化手法として音源へのノイズ付加を利用し、ユーザの聞きやすさを損なうことなく、電磁波を通じた情報復元の困難にする対策手法の開発を目的とする。本発表では、音源に逆位相のノイズを付加した2つの音声をステレオ機能を用いることで、聞き手にはノイズが打ち消される情報漏えい対策手法について検討した。 | |||||||
| 山﨑 心 | M, 2回目発表 | ディペンダブルシステム学 | 井上 美智子 | 笠原 正治 | 江口 僚太 | 笹田 大翔 |
|
tle:Loosely-Stabilizing Ranking in Population Protocols with an Upper Bound on Population Size
abstract:In recent years, autonomous distributed systems consisting of low-performance devices, such as molecular robots and sensor networks, have gained significant attention. The population protocol model has been proposed as an abstraction of such systems. One key issue in the population protocol model is the ranking problem, which involves assigning unique and consecutive integer ranks starting from 0 to anonymous devices. By solving the ranking problem through the cooperative behavior of devices, task allocation becomes possible, enabling efficient utilization of groups of low-performance devices. However, it is known that in this model, if devices do not know the exact population size, \(n\), it is impossible to design a self-stabilizing ranking protocol that starts from any initial state and gets and keeps the ranking. In this study, we propose a loosely-stabilizing ranking protocol that relaxes the strict requirement of stability for self-stabilization. language of the presentation: Japanese 発表題目:個体群プロトコルにおける個体数の上界を用いた緩安定ランキング 発表概要:近年，分子ロボットやセンサネットワークのような低性能デバイスから構成される自律分散システムが注目を集めており，これら自律分散システムの抽象モデルとして個体群プロトコルモデルが提案されている．個体群プロトコルモデルにおける重要な問題として，ランキング問題が存在する．ランキング問題とは，匿名のデバイスにユニークかつ0から連続した整数のランクを割り当てる問題である．デバイス同士の協調動作によってランキング問題を解決達成することにより，タスク分割が可能となり，低性能デバイス群を効率よく活用できる．しかしながらこのモデルにおいて，個体が正確な個体群のサイズ$n$を知らない場合，任意の初期状態からランキングを終える自己安定ランキングプロトコルが存在しないことが知られている．そこで本研究では，自己安定性の安定性の要求を緩和した緩安定ランキングを提案する． | ||||||
| 安藤 大輝 | M, 2回目発表 | 大規模システム管理 | 笠原 正治 | 井上 美智子 | 中畑 裕 | |
|
title: Computing Betweenness Centrality in Time-varying Network using Zero-suppressed Binary Decision Diagram
abstract: Betweenness Centrality (BC) is a measure of the importance of the vertices in a network. BC has been extensively studied for a variety of static networks whose topologies don't change over time. However, real-world networks are rarely static, we can consider BC in time-varying networks as well. BC value of a vertex quantifies how many shortest paths go through this vertex. Thus, BC computation is closely related to counting shortest paths. Four different shortest paths can be defined in time-varying networks, while some of four shortest paths can be enumerated in polynomial time, others are computationally hard (#P-hard). We devise a method of BC computation for time-varying networks using Zero-suppressed binary Decision Diagrams (ZDD). In several experiments, we compare the construction time of ZDD and sizes of ZDDs. language of the presentation: Japanese 発表題目: ZDDを用いた時間変化するネットワークに対する媒介中心性計算 発表概要: ネットワークの頂点の重要性を測る尺度に媒介中心性がある．従来，媒介中心性はネットワーク構造が時間変化しない，静的なネットワークを対象としてきた．一方，実世界のネットワークが静的であることは稀であり，時間変化するネットワークの媒介中心性についても研究されている．媒介中心性の計算には全最短パスの数え上げを要する．ネットワークの時間変化を考慮した結果，4つの最短の定義が生まれるが，それらの最短パスの列挙には#P困難に分類されるケースが存在する．そこで我々は，ZDDを用いた最短パス列挙を基に，媒介中心性を効率良く計算する手法を提案する．また，計算機実験により，各最短パスに対する媒介中心性の計算時間およびZDDサイズを比較する． | ||||||
| 久睦 竜主 | M, 2回目発表 | 大規模システム管理 | 笠原 正治 | 藤川 和利 | 原 崇徳 | |
|
title: On Integer Linear Programming Formulation for Energy Efficient Resource Allocation in an Open Radio Access Network
abstract: As Open Radio Access Network (O-RAN) architecture advances, the RAN Intelligent Controller (RIC) emerges as a critical component in managing dynamic network slicing. However, the increasing demands of slice mobility have raised concerns about the energy consumption required to maintain efficient network performance. In this presentation, we formulate an energy-efficient strategy as an integer linear programming (ILP) problem to optimize resource allocation in the RIC. Through simulations, we demonstrate the performance limitation of the proposed ILP-based resource allocation under service quality constraints, offering a roadmap for energy-efficient O-RAN management. language of the presentation: Japanese | ||||||
| 田口 穂鷹 | M, 2回目発表 | 大規模システム管理 | 笠原 正治 | 藤川 和利 | 原 崇徳 | |
|
title: On Real-Time Intrusion Detection System Using Autoencoder and Extended Berkeley Packet Filter
abstract: Real-time intrusion detection systems (IDSs) require not only efficiently collecting traffic but also ensuring security level. The existing work has addressed this issue by offloading the IDS with machine learning to eBPF under the eBPF constraints imposed to ensure kernel safety and stability. However, there have still been challenges in real-time packet labeling and accurate inference for unknown traffic trends. In this presentation, we propose a real-time IDS based on eBPF that is adaptive to changes in traffic trends by focusing on an autoencoder, which is one of unsupervised learning. More specifically, the proposed scheme only collects packets included in a whitelist, which are regarded as benign, in the kernel space using eBPF. After the collected packets are sent to the user space, the autoencoder is trained using quantization-aware training (QAT). The parameters of the trained model are quantized to an integer type and passed through BPF maps to the IDS running in the kernel space. In the kernel space, the trained model performs the inference by checking the difference between the input and output of the autoencoder. This series of processes links the learning and inference processes, enabling the realization of the real-time IDS that can dynamically adapt to unknown traffic trends. Through experiments, we demonstrate the potential of the proposed IDS in terms of throughput and inference performance. language of the presentation: Japanese | ||||||