• Journal of KIISE:Information Networking
• /
• v.30 no.2
• /
• pp.233-243
• /
• 2003

With the original Transmission Control Protocol(TCP) design, which is particularly targeted at the wired networks, a packet loss is assumed to be caused by the network congestion. In the wireless environment where the chances to lose packets due to transmission bit errors are not negligible, though, this assumption may result in unnecessary TCP performance degradation. In this paper, we propose three schemes that improve the ability to conceal the packet losses in the wireless network while limiting the degree of violating TCP end-to-end semantics to a temporary incidents. If there happens a packet loss at the wireless link and there is a chance that the loss is noticed by the sending TCP, the proposed schemes send an indirect acknowledgement. Each of the proposed schemes uses different criteria to decide whether there is a chance that the packet loss occurred in the wireless part is noticed by the sender. In order to limit the buffer overhead in the base, the indirect acknowledgements are issued only when the length of buffer is less than a certain threshold. We use simulation to compare the overhead and the performance of the proposed schemes, and to show that the proposed schemes improve the TCP performance compared to Snoop with a limited amount of buffer at the base station.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2007.05a
• /
• pp.1356-1358
• /
• 2007

데이터 병합은 무선 센서 네트워크에서 발생하는 통신량 축소로 에너지 소비의 최소화를 달성하기 위해 적용되는 기법이다. 이런 데이터 병합의 시간 제어는 종단간 지연시간을 최적화하면서 병합 시간의 적응적인 조절로 WSNs의 성능을 크게 향상시킬 수 있다. 따라서 본 논문은 WSNs에서 데이터 정확성과 종단간 지연시간 사이에 발생하는 상충조건의 최적화를 위해 무선 손실을 고려한 적응성 있는 타이밍 제어 프로토콜을 설계하였다. 이 프로토콜은 무선 손실의 변화에 적응성을 있는 데이터 병합을 수행할 수 있도록 선택적 참여율과 중요도를 결정함으로써 네트워크 전체에 패킷 전송 횟수를 균등하게 분배하여 전체 네트워크 안정성을 높일 수 있도록 하였다.

• Proceedings of the Korea Multimedia Society Conference
• /
• 2002.11b
• /
• pp.644-648
• /
• 2002

멀티미디어 서비스를 제공하는 미래의 이동통신 시스템에서는 데이터의 신뢰성 있는 전송을 위해 데이터링크 계층에서 제공하는 프레임 에러율(FER: Frame Error Rates)을 감소시켜야 한다. 이를 위해 라디오링크 프로토콜(RLP: radio link protocol) 계층에서는 ARQ(Automatic Repeat request) 기법을 사용한다. 본 논문에서는 기지국에서 이동 호스트로 전송되는 국부적인 재전송 프레임(Frame)의 순서가 연속된 에러로 인하여 순서가 뒤바뀌고, 이로 인하여 기지국 데이터 링크계층에서 잦은 재전송으로 발생되는 높은 상호지연을 줄이기 위해 개선된 비순서적 재전송 기법을 제안하였다. 제안된 기법은 재전송방법 중 선택적 거절(selective-reject) ARQ의 비순서적 재전송에 수정한 빠른 재전승 기법을 적용하였으며, 기지국과 이동호스트 사이에서 데이터그램(Datagram)의 단편화(fragment)로 인한 링크계층의 Re-Sequencing을 하지 않기 때문에, 종단간 지연된 성능을 향상 시킬 수 있다.

• Journal of Korea Multimedia Society
• /
• v.12 no.11
• /
• pp.1593-1608
• /
• 2009

If a certain relay node in multi-hop wireless networks might become a malicious node that does not cooperate with other nodes or a selfish node, network throughput will be dramatically decreased. Most of existing ad hoc routing protocols assuming that the nodes will fully cooperate with other nodes do not resolve the problem of network performance degradation due to malicious and selfish nodes. This paper presents the CARE (Cooperative Ad hoc routing protocol based REputation) scheme incorporating the reputation management that can achieve a multi-hop wireless network with high throughput performance. The proposed scheme provides the horizontal cross-layer approach which can identify misbehaving malicious, selfish nodes dropped out of the hop-by-hop based packet processing in the network and then set up an optimal packet routing path that will detour misbehaving nodes. And the vertical cross-layer approach contained in the CARE scheme attempts to improve the quality of routing paths by exploiting the quality of link information received from the MAC layer. Besides, it provides high TCP throughput by exploiting the reputation values of nodes acquired from the network layer into the transport layer. A case study on experiments and simulations shows that the CARE scheme incorporating vertical and horizontal cross-layer approaches yields better performance in terms of the low rate of packet loss, fast average packet delivery time, and high TCP throughput between end-to-end nodes.

• The Journal of the Acoustical Society of Korea
• /
• v.39 no.1
• /
• pp.24-31
• /
• 2020

In this paper, the deep convolutional neural network for sound event detection is studied. Especially, the end-to-end neural network, which generates the detection results from the input audio waveform, is studied for weakly supervised problem that includes weakly-labeled and unlabeled dataset. The proposed system is based on the network structure that consists of deeply-stacked 1-dimensional convolutional neural networks, and enhanced by the skip connection and gating mechanism. Additionally, the proposed system is enhanced by the sound event detection and post processings, and the training step using the mean-teacher model is added to deal with the weakly supervised data. The proposed system was evaluated by the Detection and Classification of Acoustic Scenes and Events (DCASE) 2019 Task 4 dataset, and the result shows that the proposed system has F₁-scores of 54 % (segment-based) and 32 % (event-based).

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.30 no.10B
• /
• pp.630-638
• /
• 2005

As real-time media applications based on IP multicast networks spread widely, the end-to-end QoS (quality of service) provisioning for these applications have become very important. To guarantee the end-to-end QoS of multi-party media applications, it is essential to monitor the time-varying status of both network metrics (i.e., delay, jitter and loss) and system metrics (i.e., CPU and memory utilization). In this paper, targeting the multicast-enabled AG (Access Grid) group collaboration tool based on multi-Party real-time media services, a hybrid monitoring scheme that can monitor the status of both multicast network and node system is investigated. It combines active monitoring and passive monitoring approaches to measure multicast network. The active monitoring measures network-layer metrics (i.e., network condition) with probe packets while the passive monitoring checks application-layer metrics (i.e., user traffic condition by analyzing RTCP packets). In addition, it measures node system metrics from system API. By comparing these hybrid results, we attempt to pinpoint the causes of performance degradation and explore corresponding reactions to improve the end-to-end performance. The experimental results show that the proposed hybrid monitoring can provide useful information to coordinate the performance improvement of multi-party real-time media applications.

• The KIPS Transactions:PartC
• /
• v.13C no.3 s.106
• /
• pp.345-352
• /
• 2006

In the integrated wired/wireless network which consists of the Internet and moving networks, multiple wireless links are used to connect a fixed host(FH) in the Internet to a mobile host(MH) in the moving network. For use in such an environment, we propose a scheme to overcome the TCP performance degradation due to the packet losses over the wireless links without losing the end-to-end TCP semantics. The proposed scheme in each mobile router(MR) allows to obtain the information regarding packet losses over the upstream wireless links based on the received packet sequence number and the ACK number. This information is delivered to the upstream router, which enables the upstream access router(AR) or MR to quickly retransmit the lost packets. The proposed scheme has the feature to quickly recover the packet losses incurred over the upstream wireless links and the performance of the proposed scheme is evaluated through simulation. It is shown that the significant performance gain can be obtained using the proposed scheme compared with the snoop mechanism which maintains end-to-end TCP semantics and does not require any additional features at the source and/or destination nodes.

• Proceedings of the Korean Institute of Communication Sciences Conference
• /
• 2006.07a
• /
• pp.674-674
• /
• 2006

• Phonetics and Speech Sciences
• /
• v.11 no.1
• /
• pp.41-49
• /
• 2019

An end-to-end speech recognition model consisting of a single integrated neural network model was recently proposed. The end-to-end model does not need several training steps, and its structure is easy to understand. However, it is difficult to understand how the model recognizes speech internally. In this paper, we visualized and analyzed the attention-based end-to-end model to elucidate its internal mechanisms. We compared the acoustic model of the BLSTM-HMM hybrid model with the encoder of the end-to-end model, and visualized them using t-SNE to examine the difference between neural network layers. As a result, we were able to delineate the difference between the acoustic model and the end-to-end model encoder. Additionally, we analyzed the decoder of the end-to-end model from a language model perspective. Finally, we found that improving end-to-end model decoder is necessary to yield higher performance.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.40 no.10
• /
• pp.2026-2034
• /
• 2015

Wireless sensor networks (WSNs) consist of a lot of sensor nodes having limited transmission range. So multi-hop transmission is used for communication among nodes but the multi-hop transmission degrade the end-to-end reliability. Multipath routing and opportunistic routing are typical approaches for guaranteeing end-to-end reliability in WSNs. The existing protocols improve the reliability effectively in small networks but they suffer from rapid performance degradation in large networks. In this paper, we propose the opportunistic multipath routing protocol for guaranteeing end-to-end reliability in large WSNs. Applying multipath routing and opportunistic routing simultaneously is very hard because their conflicting routing features. The proposed protocol applies these approaches simultaneously by section-based routing thereby enhancing end-to-end reliability. Additionally, the proposed protocol guarantees required reliability by the concept of section reliability. The section reliability over a certain level might satisfy required end-to-end reliability. Our simulation results show that the proposed protocol is more suitable for guaranteeing reliability than existing protocols in large-scale WSNs.