• 한국정보처리학회:학술대회논문집
• /
• 한국정보처리학회 2019년도 추계학술발표대회
• /
• pp.211-212
• /
• 2019

Slicing Radio Access Network (RAN) can help in effectively utilizing the network bandwidth and to better manage the increasing traffic over interent. RAN slicing system discussed in this paper is based on an open-source slicing mechanism in which we write a JSON configuration file for slicing policy and send it to the FlexRAN controller. FlexRAN controlls the core networks (CNs) through OAI-RAN on the evolved packet core (EPC) component of this system. Each CN is responsible for handling a saperate RAN slice. The type of internet traffic is identified by the FlexRAN crontroller and is sent to the respective CN through OAI-RAN. CN handles the traffic according to the allocated bandwidth and in this way the internet traffic is sliced inside the EPC component.

• Journal of information and communication convergence engineering
• /
• 제18권1호
• /
• pp.1-7
• /
• 2020

This paper proposes a fast-processing and low-cost hardware multiple input multiple output (MIMO) channel emulator. The channel emulator is an important component of hardware-based simulation systems. The novelty of this work is the use of sharing and pipelining functions to reduce hardware resource utilization while maintaining a high sample rate. In our proposed emulator, the samples are created sequentially and interpolated to ensure the sample rate is equal to the base band rate. The proposed 4 × 4 MIMO requires low-cost hardware resource so that it can be implemented on a single field-programmable gate array (FPGA) chip. An implementation on Xilinx Virtex-7 VX980T was found to occupy 10.47% of the available configurable slice registers and 12.58% of the FPGA's slice lookup tables. The maximum frequency of the proposed emulator is 758.064 MHz, so up to 560 different paths can be processed simultaneously to generate 560 × 758 million × 2 × 32 bit complex-valued fading samples per second.

• 방송공학회논문지
• /
• 제7권4호
• /
• pp.335-344
• /
• 2002

관심 영역 부호화 방법은 관심 영역을 고화질로 부호화하고 배경 영역을 많이 압축함으로써 주관적 화질을 향상시키는 방법이다. 본 논문에서는 관심 영역 부호화를 효율적으로 수행할 수 있는 새로운 슬라이스 구조인 내부 영상 슬라이스 구조를 제안한다. 또한 제안한 내부 영상 슬라이스 구조를 사용할 때에 관심 영역과 배경 영역에 비트율을 할당하는 방법에 대해 다루었다. 비트율을 할당할 때 관심 영역과 배경 영역의 양자화 파라메터의 간격을 고정시켜 빠르게 양자화 파라메터를 결정하도록 하고, 특히 각 영역간에 경계가 드러나지 않도록 화질이 점차적으로 변하도록 양자화 파라메터를 설정하였다. 또한 오류가 있는 전송 환경에서 관심 영역을 배경 영역보다 오류로부터 더 많이 보호하여 주관적 화질을 향상시켰다.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제11권7호
• /
• pp.3309-3328
• /
• 2017

The application of Internet of Things (IoT) in the next generation cellular networks imposes a new characteristic on the data traffic, where a massive number of small packets need to be transmitted. In addition, some emerging IoT-based emergency services require a real-time data delivery within a few milliseconds, referring to as ultra-low latency transmission. However, current techniques cannot provide such a low latency in combination with a mice-flow traffic. In this paper, we propose a dynamic resource reservation schema based on an air-interface slicing scheme in the context of a massive number of sensors with emergency flows. The proposed schema can achieve an air-interface latency of a few milliseconds by means of allowing emergency flows to be transported through a dedicated radio connection with guaranteed network resources. In order to schedule the delay-sensitive flows immediately, dynamic resource updating, silence-probability based collision avoidance, and window-based re-transmission are introduced to combine with the frame-slotted Aloha protocol. To evaluate performance of the proposed schema, a probabilistic model is provided to derive the analytical results, which are compared with the numerical results from Monte-Carlo simulations.

• 한국발생생물학회지:발생과생식
• /
• 제16권4호
• /
• pp.265-270
• /
• 2012

Ethanol actions in the amygdala formation may underlie in part the reinforcing effects of ethanol consumption. Previously a physiological phenomenon in the basolateral amygdala (BLA) that is dependent on neuronal network activity, compound postsynaptic potentials (cPSPs) were characterized. Effects of acute ethanol application on the frequency of cPSPs were subsequently investigated. Whole cell patch clamp recordings were performed from identified projection neurons in a rat brain slice preparation containing the amygdala formation. Acute ethanol exposure had complex effects on cPSP frequency, with both increases and decreases dependent on concentration, duration of exposure and age of the animal. Ethanol produces complex biphasic effects on synaptically-driven network activity in the BLA. These findings may relate to subjective effects of ethanol on arousal and anxiolysis in humans.

• 한국멀티미디어학회논문지
• /
• 제22권5호
• /
• pp.573-579
• /
• 2019

Alberta Stroke Program Early Computed Tomographic Scoring (ASPECTS) is a 10-point CT-scan score designed to quantify early ischemic changes in patients with acute ischemic stroke. However, an assessment of ASPECTS remains a challenge for neuroradiologists in stroke centers. The purpose of this study is to develop an automated ASPECTS scoring system that provides decision-making support by utilizing binary classification with three-dimensional convolutional neural network to analyze CT images. The proposed method consists of three main steps: slice filtering, contrast enhancement and image classification. The experiments show that the obtained results are very promising.

• Korean Journal of Radiology
• /
• 제22권3호
• /
• pp.476-488
• /
• 2021

Objective: We aimed to develop a deep neural network for segmenting lung parenchyma with extensive pathological conditions on non-contrast chest computed tomography (CT) images. Materials and Methods: Thin-section non-contrast chest CT images from 203 patients (115 males, 88 females; age range, 31-89 years) between January 2017 and May 2017 were included in the study, of which 150 cases had extensive lung parenchymal disease involving more than 40% of the parenchymal area. Parenchymal diseases included interstitial lung disease (ILD), emphysema, nontuberculous mycobacterial lung disease, tuberculous destroyed lung, pneumonia, lung cancer, and other diseases. Five experienced radiologists manually drew the margin of the lungs, slice by slice, on CT images. The dataset used to develop the network consisted of 157 cases for training, 20 cases for development, and 26 cases for internal validation. Two-dimensional (2D) U-Net and three-dimensional (3D) U-Net models were used for the task. The network was trained to segment the lung parenchyma as a whole and segment the right and left lung separately. The University Hospitals of Geneva ILD dataset, which contained high-resolution CT images of ILD, was used for external validation. Results: The Dice similarity coefficients for internal validation were 99.6 ± 0.3% (2D U-Net whole lung model), 99.5 ± 0.3% (2D U-Net separate lung model), 99.4 ± 0.5% (3D U-Net whole lung model), and 99.4 ± 0.5% (3D U-Net separate lung model). The Dice similarity coefficients for the external validation dataset were 98.4 ± 1.0% (2D U-Net whole lung model) and 98.4 ± 1.0% (2D U-Net separate lung model). In 31 cases, where the extent of ILD was larger than 75% of the lung parenchymal area, the Dice similarity coefficients were 97.9 ± 1.3% (2D U-Net whole lung model) and 98.0 ± 1.2% (2D U-Net separate lung model). Conclusion: The deep neural network achieved excellent performance in automatically delineating the boundaries of lung parenchyma with extensive pathological conditions on non-contrast chest CT images.

• 방송공학회논문지
• /
• 제22권4호
• /
• pp.484-495
• /
• 2017

최근 디스플레이 기기의 발전과 기가 네트워크 등의 전송 대역폭 확대로 인해 대형 파노라마 영상, 4K Ultra High-Definition 방송, Ultra-Wide Viewing 영상 등 2K 이상의 초고해상도 영상의 수요가 폭발적으로 증가하고 있다. 이러한 초고해상도 영상은 데이터양이 매우 많기 때문에 부호화 효율이 가장 높은 High Efficiency Video Coding(HEVC) 비디오 부호화 표준을 사용하는 추세이다. HEVC는 가장 최신의 비디오 부호화 표준으로 다양한 부호화 툴을 이용하여 높은 부호화 효율을 제공하지만 복잡도 또한 이전 부호화 표준과 비교하여 매우 높다. 특히 초고해상도 영상을 HEVC 복호기로 실시간 복호화 하는 것은 매우 높은 복잡도를 요구한다. 따라서 본 논문에서는 고해상도 및 초고해상도 영상에 대한 HEVC 복호기의 복호화 속도를 개선시키고자 HEVC에서 지원하는 슬라이스(Slice)와 타일(Tile) 부호화 툴을 사용하여 각 슬라이스 혹은 타일을 동시에 처리하며 디블록킹 필터 과정에서도 소정의 블록 크기만큼 동시에 처리하는 데이터-레벨 병렬 처리 방법을 소개한다. 이는 독립 복호화가 가능한 타일, 슬라이스, 혹은 디블록킹 필터에서 동일 연산을 다중 스레드에 분배하는 방법으로 복호화 속도를 향상 시킬 수 있다. 실험에서 제안 방법이 HEVC 참조 소프트웨어 대비 4K 영상에 대해 최대 2.0배의 복호화 속도 개선을 얻을 수 있음을 보인다.

• 인터넷정보학회논문지
• /
• 제22권5호
• /
• pp.27-33
• /
• 2021

With the broad adoption of the Internet of Things (IoT) in a variety of scenarios and application services, management and orchestration entities require upgrading the traditional architecture and develop intelligent models with ultra-reliable methods. In a heterogeneous network environment, mission-critical IoT applications are significant to consider. With erroneous priorities and high failure rates, catastrophic losses in terms of human lives, great business assets, and privacy leakage will occur in emergent scenarios. In this paper, an efficient resource slicing scheme for optimizing federated learning in software-defined IoT (SDIoT) is proposed. The decentralized support vector regression (SVR) based controllers predict the IoT slices via packet inspection data during peak hour central congestion to achieve a time-sensitive condition. In off-peak hour intervals, a centralized deep neural networks (DNN) model is used within computation-intensive aspects on fine-grained slicing and remodified decentralized controller outputs. With known slice and prioritization, federated learning communications iteratively process through the adjusted resources by virtual network functions forwarding graph (VNFFG) descriptor set up in software-defined networking (SDN) and network functions virtualization (NFV) enabled architecture. To demonstrate the theoretical approach, Mininet emulator was conducted to evaluate between reference and proposed schemes by capturing the key Quality of Service (QoS) performance metrics.

• 정보과학회 논문지
• /
• 제43권11호
• /
• pp.1223-1232
• /
• 2016

가상 네트워크 기술은 다양한 서비스의 요구조건을 반영한 네트워크를 제공할 수 있다. 다양한 서비스의 요구조건을 반영하기 위해 효율적인 리소스 분배 기술이 필요하다. 기존의 트래픽 대역폭 분배 기법들은 다운링크 트래픽만 제어하거나 네트워크의 트래픽 상황을 고려하지 않는다. 무선 네트워크에서 다운링크와 업링크는 같은 자원을 공유한다. 또한, 기존의 트래픽 대역폭 분배 기법들은 모든 스테이션이 포화된 트래픽을 발생시킨다고 가정한다. 그래서 기존의 트래픽 대역폭 분배 기법들은 가상 무선 네트워크에서 트래픽 제어를 할 수 없다. 본 논문에서는 이러한 문제들을 해결하기 위해 트래픽 기반 대역폭 제어 기법을 제안한다. 가상 네트워크에 SDN을 적용하고 각 스테이션의 트래픽을 모니터링하고 비포화 트래픽을 발생시키는 스테이션을 탐색한다. 또한, 모니터링 정보를 기반으로 업링크와 다운링크 트래픽을 동적으로 제어한다. 실제 테스트베드 구성 후, 기존의 기법과 비교 결과, 트래픽 대역폭 분배 성능이 최대 14% 개선되었다.