• 한국멀티미디어학회:학술대회논문집
• /
• 한국멀티미디어학회 1998년도 추계학술발표논문집
• /
• pp.202-207
• /
• 1998

The network reliability is to be computed in terms of the terminal reliability. The computation of a terminal reliability is started with a Boolean sum of products expression corresponding to simple paths of the pair of nodes. This expression is then transformed into another equivalent expression to be a Disjoint Sum of Products form. But this computation of the terminal reliability obviously does not consider the communication between any other nodes but for the source and the sink. In this paper, we derive the overall network reliability which all other remaining nodes. For this, we propose a method to make the SOP disjoint for deriving the network reliability expression from the system success expression using the modified Sheinman's method. Our method includes the concept of spanning trees to find the system success function by the Cartesian products of vertex cutsets.

• 대한전자공학회:학술대회논문집
• /
• 대한전자공학회 2003년도 신호처리소사이어티 추계학술대회 논문집
• /
• pp.259-262
• /
• 2003

A efficient P2P (peer to peer) networking for sharing multimedia contents is proposed to overcome the limitations of unstructured approaches. The proposed approach is attempting to organize the participating P2P nodes by modifying a n-dimensional cartesian-coordinate space DHT (distributed hash table), CAN (content addressable network). The network identifiers (e.g., network prefix of IP address) of participating nodes are mapped into the CAN virtual coordinate space (in 2-d) and nodes with similar identifiers are grouped into the same zone. The proposed scheme is expected to show some level of concentration reflecting the network identifiers. Network simulator-based evaluation is performed to verify the effectiveness of the proposed scheme.

• 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.

• 한국멀티미디어학회논문지
• /
• 제21권3호
• /
• pp.391-399
• /
• 2018

In this paper, we construct and compare various guitar chord classification systems using perceptron neural network and convolutional neural network without pre-processing other than Fourier transform to identify the optimal chord classification system. Conventional guitar chord classification schemes use, for better feature extraction, computationally demanding pre-processing techniques such as stochastic analysis employing a hidden markov model or an acoustic data filtering and hence are burdensome for real-time chord classifications. For this reason, we construct various perceptron neural networks and convolutional neural networks that use only Fourier tranform for data pre-processing and compare them with dataset obtained by playing an electric guitar. According to our comparison, convolutional neural networks provide optimal performance considering both chord classification acurracy and fast processing time. In particular, convolutional neural networks exhibit robust performance even when only small fraction of low frequency components of the data are used.

• 전기학회논문지
• /
• 제66권3호
• /
• pp.575-580
• /
• 2017

In these days, the big-size and high resolution multimedia file is widely used through networks. To transfer and service effectively, the internet network technology is necessary to substitute broadcasting. Normally Content Delivery Network(CDN) is widely used in conventional internet for multimedia services. But it has a small bandwidth to service. So to solve this problems, many researchers have suggest the protocol for download, content distribution/saving, server synchronization, caching, pushing rate, and streaming etc. But all of these has some defects like low resolution, packets loss and delay, real application implementations etc. So, this paper suggests a new method of network protocol based on P2P collaboration and user's content using information. And it evaluated the performance of suggested method. As the results, it showed the effectiveness of 4 performances indices : download speed, decreasing rate of connected user in same time, adaptive hit ratio, traffic decreasing rate.

• 한국멀티미디어학회논문지
• /
• 제7권6호
• /
• pp.814-823
• /
• 2004

Current video games and game room games are played as a single player mode on the basis of various emulators. With the evolution of data communications and game technology, a new trend in the game industry has made the primary interests of game developers and companies in the game industry be moved toward a multiplayer mode from the traditional single player mode. In this paper, we represent how to implement a network game platform by allowing network modules to be run in conjunction with the current video emulator games. It also suggests a synchronization scheme for real-time game playout and practical mechanism that can support network games to be played with the Peer-to-Peer process using a lobby system.

• 한국멀티미디어학회논문지
• /
• 제19권8호
• /
• pp.1415-1423
• /
• 2016

Interference in device-to-device (D2D) communication underlaying cellular network needs to be elaborately investigated because of channel sharing. The objective is to improve the quality of D2D communications while maintaining high performance for cellular users. In this paper, we solve the above problem by jointly considering channel allocation and power control using coalition formation game. Our cooperative game theoric approach allows to enhance network-wide performance. We design a merge-and-split algorithm to deal with the complexity of the combinatorial structure in coalition formation problem. The analytical and numerical results show that our algorithm converges to a stable point which achieves high network performance.

• 한국멀티미디어학회논문지
• /
• 제19권9호
• /
• pp.1690-1697
• /
• 2016

In a beacon-enabled Zigbee network, the slotted CSMA/CA mechanism based on the super frame structure fairly provides communication chance for each node and makes a reasonable usage of the available energy. In the case of wireless nano sensors that are implanted into the target human body area for detecting disease symptoms or virus, such a nano-network requires a similar type of channel sharing and transmission of short length event-driven data. In this paper, for nano-network's in-body applications, we aim to design conceptually a new super frame derived from the existing beacon-enabled Zigbee MAC protocol. And we analyze the efficiency of the proposed super frame in the aspect of practical deployment.

• 한국멀티미디어학회논문지
• /
• 제23권12호
• /
• pp.1476-1485
• /
• 2020

Segmenting OCT retinal images into layers is important to diagnose and understand the progression of retinal diseases or identify potential symptoms. The task of manually identifying these layers is a difficult task that requires a lot of time and effort even for medical professionals, and therefore, various studies are being conducted to automate this using deep learning technologies. In this paper, we use cGAN-based neural network to automatically segmenting OCT retinal images into seven terrain-type regions defined by six layer boundaries. The network is composed of a Segnet-based generator model and a discriminator model. We also proposed a dynamic programming algorithm for refining the outputs of the network. We performed experiments using public OCT image data set and compared its performance with the Segnet-only version of the network. The experimental results show that the cGAN-based network outperforms Segnet-only version.

• 한국멀티미디어학회논문지
• /
• 제25권1호
• /
• pp.72-81
• /
• 2022

Visual Object Tracking is known as the most fundamental problem in the field of computer vision. Object tracking localize the region of target object with bounding box in the video. In this paper, a custom CNN is created to extract object feature that has strong and various information. This network was constructed as a Siamese network for use as a feature extractor. The input images are passed convolution block composed of a bottleneck layers, and features are emphasized. The feature map of the target object and the search area, extracted from the Siamese network, was input as a local proposal network. Estimate the object area using the feature map. The performance of the tracking algorithm was evaluated using the OTB2013 dataset. Success Plot and Precision Plot were used as evaluation matrix. As a result of the experiment, 0.611 in Success Plot and 0.831 in Precision Plot were achieved.