| JIA HAOHUI | D, 中間発表 | ネットワークシステム学 | 岡田 実, 笠原 正治, 東野 武史, DUONG QUANG THANG, Chen Na |
|
title: Reliable Deep Learning Integrated Beam Prediction for mmWave Massive MIMO System
abstract: The deep learning (DL) driven massive multiple input multiple output (massive MIMO) beam prediction is outperforming but sensitive with noise. In order to overcome the noise and the millimeter wave (mmWave) channel fading attenuation, we propose a sparse feature driven vision transformer (SF-ViT) DL model which can non-deterministically quantify the channel signals as the binary sequences and extract the spatial characteristics from antenna indices. Specifically, we track the peaks of signal as essential components and further purify the sparse sequence by dynamic thresholds to remove the noise effects. Due to the spatial coherence of massive MIMO systems, we apply the self-attention mechanism to explore the spatial characteristics from the sparse features of antenna indices. Moreover, we consider a wise-scoring scheme to further improve the robustness of inference and obtain the reliable predictions underlying the low signal-to-noise ratio (SNR) level. language of the presentation: English | |||
| WANG ZAOSHI | M, 1回目発表 | ネットワークシステム学 | 岡田 実, 林 優一, 東野 武史, DUONG QUANG THANG |
|
title:Deep Learning-Based variable scaling Millimeter-Wave Beam Training abstract: Due to the attenuation of millimeter waves, the requirements for beamforming in communication systems are getting higher and higher. The overhead of traditional beam scanning methods in massive MIMO systems is huge. Deep learning (DL) techniques have been shown to perform better in massive MIMO systems relative to traditional beam selection methods. In order to improve the feature extraction of beam by neural network, we hope to extract channel features at different scales by designing an inception-based DL model. Our goal is to train and predict the corresponding narrower waves with higher beamfoeming gain based on the smaller number of wider waves with larger coverage in the same space. First, we need to generate a dataset of channel characteristics in a simulated environment, and then classify it into wide-wave and narrow-wave through the codebook. in the same space. In addition, we will also consider different methods such as diagonal elements to further reduce the computational overhead of training data. language of the presentation: English | |||
| 北村 圭輝 | M, 1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, Youngwoo Kim |
|
title: Efficient Evaluation Methods for Zoning Countermeasures Based on Separation of Leakage Processes
abstract: Zoning is a countermeasure against TEMPEST, in which an attacker obtains information through electromagnetic waves emanated from information equipment. Zoning is achieved by combining various countermeasure technologies, but it requires a significant amount of time for trial and error. Therefore, this study proposes a method to separate the EM information leakage process into leakage source, propagation loss, and background noise and efficiently evaluate countermeasure techniques' effectiveness. language of the presentation: Japanese 発表題目: 漏えい過程の分離に基づくゾーニング対策の効率的な評価法の検討 発表概要: 情報機器から放射される電磁波を通じて攻撃者が情報を取得するTEMPESTの対策としてゾーニングがある。ゾーニングは様々な対策技術を組み合わせて実現するが、試行錯誤に多大な時間を要する。そこで本稿では、電磁波による情報漏えいを、漏えい源、伝搬損失、背景雑音に分離し、対策技術の効果を効率的に評価するための手法を提案する。 | |||
| 近藤 嵩之 | M, 1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, Youngwoo Kim |
|
title: Threat Analysis of a Data Injection Attack Against CAN BUS with Malicious Modification
abstract: The CAN BUS is a key network component that interconnects electronic control units (ECUs). If the CAN BUS is attacked, the security of vehicles could be significantly compromised. This study analyzes the security threats posed by malicious modifications to the CAN BUS in the automotive supply chain. language of the presentation: Japanese 発表題目: 悪意ある改変が行われたCAN BUSへのデータ注入攻撃の脅威分析 発表概要: CAN BUSは電子制御装置（ECU）を相互に接続するネットワークの要であり、CAN BUSが攻撃された際には自動車のセキュリティが大幅に低下する恐れがある。本研究では、自動車を生産するサプライチェーンにおいて、CAN BUSに悪意ある改造がなされた場合に生ずるセキュリティの脅威について分析する。 | |||
| 水黒 知也 | M, 1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, Youngwoo Kim |
|
title: Assessment of the Influence of Printed Circuit Board Design on Information Leakage from Cryptographic Devices
abstract: It is known that the obtainability of information leaked from a cryptographic module through electromagnetic (EM) waves depends on the physical structure of the device in which the cryptographic module is mounted. It has been reported that the size of the printed circuit board (PCB) and the length and number of lines connected to the device affect this. On the other hand, the design of PCBs also varies depending on the devices and may affect the information acquisition and the size of PCBs and the lines connected to the device. This research focuses on AES, a widely used cryptographic algorithm. We show that the design of PCBs on which the cryptographic module is implemented affects the amplitude and spectrum of the leaked EM wave, which affects the obtainability of the secret key. language of the presentation: Japanese 発表題目: 基板のデザインが暗号デバイスからの情報漏えいに与える影響の評価 発表概要: 暗号モジュールから電磁波を通じて漏えいする情報の取得性は、暗号モジュールが実装される機器の物理構造に依存することが知られている。これまで、機器を構成するプリント基板のサイズや機器に接続される線路の長さや本数が影響することが報告されている。一方、基板のデザインも機器により様々であり、プリント基板のサイズや接続線路と同様に情報の取得性に影響を与える可能性がある。本稿では、広く利用されている暗号アルゴリズムであるAESに着目し、暗号モジュールが実装されたプリント基板のデザインが、漏えい電磁波の振幅やスペクトルに変化を及ぼし、最終的に秘密鍵の取得性に影響を与えることを明らかにする。 | |||