• 한국멀티미디어학회논문지
• /
• 제18권4호
• /
• pp.437-442
• /
• 2015

Voice activity detection is very important process that voice activity separated form noisy speech signal for speech enhance. Over the past few years, many studies have been made on voice activity detection, but it has poor performance in low signal to noise ratio environment or fickle noise such as car noise. In this paper, it proposed new voice activity detection algorithm using ensemble variance based on wavelet band entropy and soft thresholding method. We conduct a survey in a lot of signal to noise ratio environment of car noise to evaluate performance of the proposed algorithm and confirmed performance of the proposed algorithm.

• 한국멀티미디어학회논문지
• /
• 제2권4호
• /
• pp.367-377
• /
• 1999

최근 들어 신속한 트랜잭션 처리를 요구하는 데이터베이스 응용이 확대되고 있다. 트랜잭션 처리 시스템에서 높은 성능을 달성하기 위한 한가지 방법으로 데이터베이스를 디스크가 아닌 주기억장치에 모두 상주시키는 것이다. 반도체 메모리의 집적도가 증가하고 가격이 하락함에 따라 모든 데이터베이스를 주기억장치에 상주시켜 트랜잭션 처리율을 높이기 위한 연구가 이루어지고 있다. 본 논문에서는 주기억장치 데이터베이스를 기반으로 한 고성능 트랜잭션 처리 시스템을 구현하기 위하여 새로운 병행수행 제어 기법과 회복 기법 그리고 저장 구조를 제안하며, 트랜잭션의 처리량과 응답속도로 평가되는 트랜잭션 처리 시스템 성능의 개선을 목적으로 한다.

• 한국멀티미디어학회논문지
• /
• 제21권10호
• /
• pp.1142-1149
• /
• 2018

Recently, computer vision application is increasing by using CNN which is one of the deep learning algorithms. However, CNN does not provide perfect classification performance due to gradient vanishing problem. Most of CNN algorithms use an activation function called ReLU to mitigate the gradient vanishing problem. In this study, four activation functions that can replace ReLU were applied to four different structural networks. Experimental results show that ReLU has the lowest performance in accuracy, loss rate, and speed of initial learning convergence from 20 experiments. It is concluded that the optimal activation function varied from network to network but the four activation functions were higher than ReLU.

• Journal of Multimedia Information System
• /
• 제7권2호
• /
• pp.137-140
• /
• 2020

In this paper, we propose a simplified tag collection algorithm to improve the performance of ISO / IEC 18000-7, the standard of active RFID systems. In the proposed algorithm, the collection command is modified to include the result of the listening period response from the previous round. The tag, which has received the collection command, checks whether the slot to which it has responded is collided, transmits additional data to its data slot without a point-to-point read command and sleep command, and transitions to the sleep mode. The collection round in the standard consists of a series of collection commands, collection responses, read commands, read responses, and sleep commands. On the other hand, in the proposed tag collection algorithm, one collection round consists only of a collection command and a collection response. As a result of performance analysis, it can be seen that the proposed technique shows superior performance compared to the standard.

• 인터넷정보학회논문지
• /
• 제2권5호
• /
• pp.93-104
• /
• 2001

본 논문에서는 멀티미디어 환경에서 La치 성능향상을 위하여 ATM-LAN으로 구성하여 토폴로지에 따라 네트워크를 분석한다. 노드 변화에 따라 스타, 링, 메쉬 토폴로지에서 Email, FTP, Http 서버의 작업처리 시간을 분석한다. 토폴로지에 따라 Email 송, 수신율의 변화와 Email, FTP, Http서버의 부하량을 분석한다. 본 논문에서 구성한 네트워크는 단지 어떤 토폴로지에서 서버에 걸리는 부하를 최소로 할 수 있는가를 고찰할 수 있었다.

• 한국통신학회논문지
• /
• 제22권11호
• /
• pp.2521-2530
• /
• 1997

In this paepr, we present a video-on-demand (VOD) system design model that address and integrates a number of inter-related issues. Then with analysis and performance evaluation, we investigate various aspects of disk and buffer managements in the given model. Based on the analysis results, we suggest that a distributed buffering scheme with intermediate buffers may te useful to transform bursty disk accesses into a continuous stream for for glitch-free performance of VOD systems. Also, through simulation, we illustrate that massive multimedia information storage design techniques such as prefetching, clustered striping, and real-time disk scheduling integrated with the distributed buffering mechanism may enhance end-to-end real-time performance of VOD systems under wide-area networks.

• 한국통신학회논문지
• /
• 제32권1B호
• /
• pp.17-21
• /
• 2007

본 논문에서는 멀티미디어 네트워크상에서 우선순위가 높은 서비스의 QoS를 보장하며 동시에 프로세서의 에너지 효율을 최대화하는 적응적 온라인 프로세서 관리기법을 제안하였다. 이 기법은 현재 네트워크의 트래픽 상황을 기반으로 하여 실시간으로 프로세서를 관리한다. 제안된 방법의 가장 큰 특징은 상호 상충하는 다양한 성능 메트릭들 사이에 적정한 균형을 이루어 안정적으로 시스템을 운영하는데 있다. 컴퓨터 시뮬레이션을 통하여 기존에 제안된 프로세서 관리기법에 비해 본 논문에서 제안된 방법이 다양한 트래픽 상황에서 우수한 성능을 가지는 것을 확인할 수 있었다.

• 한국게임학회 논문지
• /
• 제19권6호
• /
• pp.49-60
• /
• 2019

Interior Mapping은 효율적으로 그래픽 리소스를 최적화하는 방법들 중에 하나로 사용되어왔다. 본 논문에서는 보편화 된 게임엔진인 Unreal Engine 4를 활용하여 Interior Mapping을 적용했을 시 3D 게임 내의 성능 변화를 검증하기 위해 실제 모델링과 Interior Mapping의 렌더링 속도(FPS), 폴리곤 개수, 쉐이더 복잡도, 각각의 리소스 용량을 비교하였다. 또한, 효율적인 기술의 적용을 위해 큐브 맵 텍스처의 해상도와 세부 정보에 따른 성능 차이를 검증하였다.

• 한국멀티미디어학회논문지
• /
• 제18권12호
• /
• pp.1468-1474
• /
• 2015

Efficient spectrum sharing is an important issue in Device-to-Device (D2D) communications underlaying cellular networks as it can mitigate the interference to cellular users and improve the performance of the systems. In this paper, we formulate the radio resource allocation in D2D communications as a mixed nonlinear integer programing. We show the formulated problem is NP-hard and thus a polynomial time algorithm to solve is not possible. Since such a problem is very hard to obtain the optimal solution within a short running time, we instead propose a fast heuristic suboptimal algorithm to mitigate the interference caused to cellular users and improve the performance of the systems. Simulation results are provided to evaluate the performance of the proposed algorithm.

• 한국멀티미디어학회논문지
• /
• 제21권4호
• /
• pp.441-449
• /
• 2018

Deep learning has been used for various processing centered on image recognition. One core algorithms of the deep learning, convolutional neural network is an deep neural network that specialized in image recognition. In this paper, we use a convolutional neural network to classify forest insects and propose an optimization method. Experiments were carried out by combining two weight initialization and six parameter update methods. As a result, the Xavier-SGD method showed the highest performance with an accuracy of 82.53% in the 12 different combinations of experiments. Through this, the latest learning algorithms, which complement the disadvantages of the previous parameter update method, we conclude that it can not lead to higher performance than existing methods in all application environments.