コロキアムB発表

日時: 6月23日(火)3限(13:30~15:00)


会場: L1

司会: Duong Quang Thang
JIA HAOHUI M, 2回目発表 ネットワークシステム学 岡田 実, 笠原 正治, 東野 武史, Dong Duong Thang, Chen Na
Title: Sparse Channel Estimation With Subcarrier Mapping Scenario For Single Carrier-FDMA
Abstract: Single carrier frequency division multiple access (SC-FDMA) system reduces a peak to average power ratio comparing with the orthogonal frequency division multiplexing (OFDM) system. And it enables to flexible frequency allocation for multiplexing user access in long-term evolution (LTE) uplink. Not only subcarrier mapping of pilot signal, but also channel estimation scheme has considerable impact for channel estimation quality. Recently, compressed sensing (CS) technique has applied for channel estimation in wireless system in order to reduce the pilot cost. However, the channel estimation quality when CS is employed for channel estimation in SC-FDMA system would be effected by the high correlation in localized frequency division multiple access scheme. This research aim to improve the accuracy with cross user data block pilots and partial information to obtain whole channel state information instead of respective estimation.
Language of the presentation: English
 
PHAM HOAI LUAN M, 2回目発表 コンピューティング・アーキテクチャ 中島 康彦, 笠原 正治, 中田 尚, Tran Thi Hong, 張 任遠
Title: High Efficient Blockchain Application and Circuit For Decentralized Network
Abstract: Blockchain technology has emerged as an adequate solution for securing cyberspace of society 5.0. Thanks to the benefits of Blockchain, IoT data are guaranteed to be secure, immutable, available, and transparent. Nevertheless, Blockchain-based current IoT applications remain low scalability and high transaction costs. In the other aspect, the mining process, adding new blocks to the chain, plays the role of the backbone to maintain IoT applications operate correctly on the Blockchain network. However, the circuit-based mining process consumes massive energy. In this thesis, we investigate the two aspects of Blockchain that are application and circuit for decentralized networks. In Blockchain application, we prefer a medical system to solve the problem of transaction costs and scalability. In the medical system, we focus on two small applications that are remote monitoring and drug management. Based on the Ethereum Blockchain platform, we propose smart contracts to manage the remote monitoring and drug management systems. Experimental results of the medical system show that transaction cost is less than 1 USD, and scalability is much higher than previous works. In the Blockchain circuit, we choose mining hardware for the most famous cryptocurrency, as called Bitcoin. The Bitcoin mining circuit is mainly composed of Double-Sha-256 calculations. To reduce power consumption, we propose a Double-SHA-256 architecture with compact message expanders. Experimental results show that the hardware efficiency of proposed Double-SHA-256 architecture is superior as compared to previous works.
Language of the presentation: English
 
MUHAMMAD RADIFAN FITRACH M, 1回目発表 大規模システム管理 笠原 正治, 安本 慶一, 笹部 昌弘, 張 元玉
Title: A Model Factor Analysis of Trust in Social Networking Service During Disaster Event
Abstract: Information and communication technology have developed rapidly and took all the users to provide unreliable information sharing inside the social networking (SNS). However, it is hard for the newcomers to distinguish the trustworthiness of the news, especially disaster-related news. To understand the process of how this information spreads, we developed a model of information dissemination towards disaster-related news in SNS. The information dissemination on SNS is modeled with both user psychology and trust model. Trough online questionnaire, we measured the cognitive evaluation of trust to understand how human trusting the news that shared by other people in SNS. Then, we conduct an agent-based modeling simulation based on trust to capture and understand how the system behave. This research aims to find the evidence of specific action performs by the user based on their knowledge and psychological side. The results of this research will help the social media provider collaboratively with government to safely provide trustable information for the users.
Language of the presentation: English
 
SRI-IESARANUSORN PANYAWUT M, 1回目発表 数理情報学 池田 和司, 安本 慶一, 吉本 潤一郎, 福嶋 誠
Title: Quantification Analysis of Behavioral Changes after Sciatic Nerve Ligation in Rats
Abstract: Measuring behavior changes in humans and animals is an ongoing research in both bioscience and neuroscience, due to its applications in nerve injury recovery. Previous study focused on sciatic nerve and its recovery process because partial damage to this nerve is leading to be paralyze symptoms of entire lower body parts. To investigate the nerve, previous experiments were conducted on rats because of the short duration of experiment. However, their data collection and analysis processes were highly time-consuming and costly, because of visual inspection-based experiment, manual labor, and complicated equipment. To deal with this issue, DeepLabCut provides easy methods for observing and recording animal behavior in diverse settings through video data. In this study, to examine the behavioral changes in rats after sciatic nerve ligation, we (i) recorded videos of 9 rats during the first and fifth day after surgery, (ii) used DeepLabCut to track the rat's positions throughout the videos, and (iii) calculated behavior indexes based on spatio-temporal patterns that measure the behavioral changes in rats. The results in this study confirm our assumption that the rat's behavior has significant changes in the experiment as indicated by the indexes. These findings motivate us to investigate the behind of recovery process, and implement real-time monitoring and analysis for measuring behavior changes in rats.
Language of the presentation: English

 
CHAUQUE EUCLIDES TELES TOMAS M, 2回目発表 情報基盤システム学 藤川 和利, 安本 慶一, 新井 イスマイル
title:Leveraging Replication to Reduce Tail Latency in Datastores
abstract: The interactive nature of modern online applications necessitates low and predictable latencies because people naturally prefer fluid response times. However, it is challenging to deliver consistent low latency since interactive online applications are typically structured as multi-tiered, distributed systems, where even serving a single end-user request (e.g., to return a web page) may involve contacting multiple servers. Significant delays at any of these servers inflate the latency observed by end users. Surveys have demonstrated that internet scale systems present long tail latency, which is detrimental for users interaction, and also for businesses. In this research a method to reduce tail latency is proposed
language of the presentation: English