• Journal of the Institute of Convergence Signal Processing
• /
• v.19 no.4
• /
• pp.162-166
• /
• 2018

Companies that use automatic feeder systems use multiple feeders in close proximity to each other for mass production. Unlike when the feeder is used independently, it does not oscillate to a value that is set to offset or freeze due to symbiosis caused by asynchronous forces hitting each other. This may result in the shipment being damaged by hitting each other or thrown to one side, preventing the transport from being moved to the next process. Whenever an unstable resonance occurs, the operator must reset the value of each feed controller or operate it several times. To address these problems, this paper proposes synchronization algorithm suitable for feed controllers using multiple units and tests are used to confirm performance improvements.

• Proceedings of the KIEE Conference
• /
• 2003.07d
• /
• pp.2193-2196
• /
• 2003

기존의 멀티레이트 입력 (MRIC) 제어기는 제어 입력신호에 채터링 현상이 발생하여 추종응답에 리플이 발생하는 문제점이 있다. 이러한 리플현상을 해결하기 위해 피드포워드 보상기를 이용하여 이런 리플현상을 제거하는 방법이 제시되기도 하였다. 본 논문에서는 상태 공간에서 현재 추종기를 기반으로 리프팅 기법을 이용한 멀티레이트 입력제어기 설계를 소개한다. 본 제어기를 사용함으로써 제어입력 신호가 채터링이 생기지않음을 이산전달함수 관점에서 최종치 정리를 이용하여 증명한다.

• Journal of IKEEE
• /
• v.28 no.1
• /
• pp.53-64
• /
• 2024

With the advancement of NLP(Natural Language Processing) models, conversational AI such as ChatGPT is becoming increasingly popular. To enhance processing speed and reduce power consumption, it is important to implement the Transformer algorithm, which forms the basis of the latest natural language processing models, in hardware. In particular, the multi-head attention and feed-forward network, which analyze the relationships between different words in a sentence through matrix multiplication, are the most computationally intensive core algorithms in the Transformer. In this paper, we propose a new variable systolic array based on the number of input words to enhance matrix multiplication speed. Quantization maintains Transformer accuracy, boosting memory efficiency and speed. For evaluation purposes, this paper verifies the clock cycles required in multi-head attention and feed-forward network and compares the performance with other multipliers.

• The Journal of the Korea Contents Association
• /
• v.15 no.10
• /
• pp.669-678
• /
• 2015

This study aims at investigating students' perceptions and peer feedback to practicum for teaching English in the English Department at a Korean university. The participants were forty-two students at an elective course, 'Method for teaching English', and the data comprised questionnaire, 12 teams of practicum, and 15 sets of PF. A 'Word Count System (WCS)' was adopted to analyze the data. The findings show that students regarded 'practicum' (52.4%) as more important than 'teacher's lectures' (42.8%), and most students (80%) applied more than 70% of lesson plans to their practicums. The practicum gave them experience of a teacher, development of confidence, recognition on their weaknesses and values of teaching. While the strengths shown in PF were mainly 'teaching methods and technique', 'use of multimedia', and 'teaching materials', the weaknesses were 'classroom interactions', 'teaching methods and techniques' and 'use of blackboard'. Overall praises were 1.8 times more than the matters which needed to be developed. The conclusion suggest that the students had their own insights toward teaching practices and how learners can be motivated.

• Proceedings of the Korea Multimedia Society Conference
• /
• 2004.05a
• /
• pp.199-202
• /
• 2004

데이터 방송은 지상파, 위성, 케이블 둥의 방송망을 이용하여 하나의 송신자가 다수의 수용자에게 디지털 멀티미디어 컨텐츠를 여러 종류의 단말기에 전송하는 것으로 컨텐츠는 프로그램에 연동되는 서비스 및 비연동형 서비스, 또는 대화형 서비스 헝태로 제공되는 다양한 서비스를 포함한다. 방송망을 이용한 데이터방송의 전송 규약은 DSM-CC에서 정의한 캐로셀(carousel) 형태로 전송된다. 캐로셀은 주기적으로 반복 전송되는 데이터 모듈을 가리키며, 캐로셀로 전송되는 데이터 중에는 MPEG-2 비디오로 인코딩되어 광고와 배경 이미지 등으로 사용할 수 있는 영상 데이터가 있다. 이 MPEG-2 비디오 영상 데이터를 TV화면에 출력하기 위해 셋톱박스의 미들웨어는 썬 마이크로시스템즈(Sun Microsystems)에서 만든 자바 기반의 JMF(Java Media Framework) 플레이어를 통해 화면에 출력한다 이렇게 제공되는 MPEG-2 비디오 영상 데이터를 비디오 드립(drips)이라 하고, 비디오 드립을 JMF 플레이어를 통해 화면에 출력하는 모드를 드립피드(drip-feed) 모드라 한다. 그러나 MHP용 셋톱박스를 위한 JMF 버전 1.0은 그대로 사용 할 수 없다. 비디오 드립 모드를 위한 구현이 없기 때문에 별도의 확장을 통해 비디오 드립을 지원하거나, 네이티브(Native) 메소드를 만들어 JMF를 확장하는 방식을 제공해야 한다. 본 논문에서는 데이터 캐로셀로 전송되는 비디오 드림 모드를 구현하기 위해 네이티브 코드로 JMF를 확장해서 비디오 드립을 JMF 플레이어를 통해 TV 화면에 출력하는 MPEG-2 비디오 드립 디코더를 설계하고 구현하였다.

• Journal of Korea Multimedia Society
• /
• v.17 no.2
• /
• pp.218-225
• /
• 2014

The marketing which uses SNS, which has the most biggest growth since appearance of Facebook, is proved to be effective, then many companies make variable attempts of marketing by using Facebook. This paper presents the method of marketing using videos, that has a major part in Facebook, and the method to raise their ratings. By using the video play button it registers information of the videos in Facebook then shares them to other people in real-time through inner News Feed. Therefore, we can maximize effect of marketing through viral marketing to use other users responses.

• The Transactions of The Korean Institute of Electrical Engineers
• /
• v.64 no.12
• /
• pp.1737-1741
• /
• 2015

A multi-channel CMOS transimpedance amplifier(TIA) array is realized in a $0.18-{\mu}m$ CMOS technology for the applications of panoramic scan LADAR systems. Each channel consists of a PIN photodiode and a feed-forward TIA that exploits an inverter input stage followed by a feed-forward common-source amplifier so as to achieve lower noise and higher gain than a conventional voltage-mode inverter TIA. Measured results demonstrate that each channel achieves $76-dB{\Omega}$ transimpedance gain, 720-MHz bandwidth, and -20.5-dBm sensitivity for $10^{-9}$ BER. Also, a single channel dissipates the power dissipation of 30 mW from a single 1.8-V supply, and shows less than -33-dB crosstalk between adjacent channels.

• Journal of Korea Multimedia Society
• /
• v.12 no.9
• /
• pp.1201-1209
• /
• 2009

Recently, the convergence of the telecommunications industry through the development of network media has brought along interactive data services in addition to the traditional broadcast services by providing direct connections between the users and the broadcasters. This can be easily achieved by the combination of existing network technology and multimedia technology, but in reality, as seen in the actual implementation of IPVT, it's difficult to provide satisfactory levels of service to users due to issues of network bandwidth and problems of streaming system. In particular, a network which is able to fully support Qos/QoE, and also a fast enough channel changing technology that is satisfactory to the user, are prerequisites for IPTV success. Therefore this thesis proposes a fast channel changing technique to delivery enhanced IPTV user experience. Recent IP set-top-boxes use only a small portion of the CPU to feed Ethernet packets to the hardware decoder and play them. Thus, by making IP set-top-boxes share cached channel information and content with each other, reducing the load on the server and enhancing the channel changing time is made possible.

• Journal of Korea Multimedia Society
• /
• v.13 no.7
• /
• pp.933-942
• /
• 2010

The convergence of the telecommunication industry through the combination of existing network technology and multimedia technology has brought along interactive data services in addition to the traditional broadcast services by providing direct connections between the users and the broadcasters. The interactive data service technology has developed a network which is able to fully support QoS and also a fast enough channel changing technology that is satisfactory to the user, are prerequisites for IPTV success, to provide satisfactory levels of service to exist users. However the current server-client system can't support all increasing needs of users. To overcome these limitations, P2P based IPTV service is currently under researched. Therefore, this thesis proposes a logical view based copy-ahead technique for P2P IPTV to improve the performance of IPTV system.

• Journal of the Institute of Electronics Engineers of Korea TC
• /
• v.42 no.11
• /
• pp.1-12
• /
• 2005

In wireless personal area networks(PANs), the successful design of channel time allocation algorithm is a key factor in guaranteeing the various quality of service(QoS) requirements for the stringent real-time constraints of multimedia services. In this paper we propose a dynamic channel time allocation algerian for providing delay guarantees to multimedia traffics such as MPEG streams in the IEEE 802.15.3 high-rate WPANs. The dynamic algorithm exploits the characteristics of MPEG stream, wherein the devices (DEVs) send their channel time requests only at the end of superframe boundaries. The algerian uses mini packets for feedback control in order to deliver dynamic parameters for channel time requests from the DEVs to the piconet coordinator (PNC). In this scheme, the duration of channel time allocated to a DEV during a superframe is changed dynamically depending on the MPEG frame type, traffic load and delay bound of the frame, etc. Performance of the proposed scheme is investigated by simulation. Our results show that compared to traditional scheme, the proposed scheme is very effective and has high performance while guaranteeing the delay bound.