http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
프랙티컬 비잔틴 장애 허용 기반의 합의 알고리즘의 평가 프레임워크
이은영(Eun-young Lee),김남령(Nam-ryeong Kim),한채림(Chae-rim Han),이일구(Il-gu Lee) 한국정보통신학회 2021 한국정보통신학회 종합학술대회 논문집 Vol.25 No.2
PBFT(Practical Byzantine Fault Tolerant)는 Pow(Proof of Work) 대비 높은 처리 속도를 보장하며 컴퓨팅 파워의 우위로 인한 기록이 번복되지 않는 절대적 최종성을 보장하는 합의 알고리즘이다. 하지만 메시지 복잡도로 인해 참여하는 노드의 수가 증가할수록 네트워크 부하가 지수적으로 증가한다는 한계가 있다. PBFT는 블록체인 네트워크의 성능을 결정짓는 중요한 요소이지만 평가지표와 평가 기술에 관한 연구는 부족한 실정이다. 본 논문에서는 PBFT를 평가할 수 있는 정량적 지표와 개선된 방안을 쉽게 평가할 수 있도록 합의 알고리즘 변경이 편리한 PBFT 평가 프레임워크를 제안한다. PBFT (Practical Byzantine Fault Tolerant) is a consensus algorithm that guarantees higher processing speed compared to POW (Proof of Work) and absolute finality that records are not overturned due to the superiority of computing power. However, due to the complexity of the message, there is a limit that the network load increases exponentially as the number of participating nodes increases. PBFT is an important factor in determining the performance of a blockchain network, but studies on evaluation metrics and evaluation technologies are lacking. In this paper, we propose a PBFT evaluation framework that is convenient to change the consensus algorithm to easily evaluate quantitative indicators and improved methods for evaluating PBFT.
딥보이스를 이용한 보이스피싱의 대응방안을 위한 음성 주파수 분석
김지후(Ji-Hu Kim),류세민(Se-min Ryu),한채림(Chae-Rim Han),이창의(Chang-Ui Lee),이종현(Jong-Hyeon Lee),이민재(Min-Jae Lee),오민규(Min-Gyu Oh),박세진(Se-Jin Park) 대한전자공학회 2023 대한전자공학회 학술대회 Vol.2023 No.11
With the development of deep learning along with artificial intelligence, ethical problems that abuse it while providing convenience to humans have increased. While the problem of voice phishing intensifies with deep voices synthesized using artificial neural networks, this paper analyzes the similarities and differences between real voices and voices coated with voice synthesis technology to explore factors that can identify voices. As a result of the paper, it was confirmed that the characteristics of the experimenters voice were maintained in certain areas even though they were synthesized as deep voices. Through this area, it was possible to specify the voice before synthesis in the deep voice.
마이크로 서비스 아키텍처 설계 및 구축을 위한 가상화 기술별 성능 비교 분석
이창의(Chang-Ui Lee),이종현(Jong-Hyeon Lee),이민재(Min-Jae Lee),오민규(Min-Gyu Oh),김지후(Ji-Hu Kim),류세민(Se-Min Ryu),한채림(Chae-Rim Han),박세진(Se-Jin Park) 대한전자공학회 2023 대한전자공학회 학술대회 Vol.2023 No.11
This paper tested servers implemented with KVM and Docker virtualization technologies in the microservice architecture. The servers implemented on top of each virtualization technology were loaded by creating threads through JMeter. We measured response time and throughput in two cases where the load on the server increases radically for one second and gradually for 30 seconds. We also tested the servers ability to recover from failures by measuring the time that the server reboots if it goes down due to a failure. For radical load tests, Docker performed approximately 1.3x better than KVM and for incremental load tests, Docker performed approximately 1.17x better than KVM. Average restart time measurements showed Docker 5.47 times faster than KVM. Through this paper, we hope to contribute to the choice of virtualization technology to developers who want to introduce micro-service architecture or adopt micro-service architecture.