• 한국음향학회지
• /
• 제39권6호
• /
• pp.505-514
• /
• 2020

심층 신경망(Deep Neural Network, DNN) 모델을 대량의 학습 데이터로 학습시키기 위해서는 많은 시간이 소요되기 때문에 병렬 학습 방법이 필요하다. DNN의 학습에는 일반적으로 Stochastic Gradient Descent(SGD) 방법이 사용되는데, SGD는 근본적으로 순차적인 처리가 필요하므로 병렬화하기 위해서는 다양한 근사(approximation) 방법을 적용하게 된다. 본 논문에서는 기존의 DNN 병렬 학습 알고리즘들을 소개하고 연산량, 통신량, 근사 방법 등을 분석한다.

• International Journal of Computer Science & Network Security
• /
• 제23권5호
• /
• pp.121-127
• /
• 2023

A 500 km/h linear motor high speed terrestrial transportation service is planned to launch 2027 in Japan. In order to support 5G service in the train, the Sub-carrier spacing frequency of 30 kHz is planned to be used instead of common 15 kHz sub-carrier spacing to mitigate Doppler effect in such high-speed transportation. In addition, to increase the cell size of 5G mobile system, plural Base Station antenna will transmit the identical Down Link (DL) signal to form the expanded cell size along the train rail. In this situation, forward and backward antenna signals will be Doppler shifted by reverse direction respectively and the receiver in the train might suffer to estimate accurate Channel Transfer Function (CTF) for its demodulation. In this paper, Delay and Doppler Profiler (DDP) based Channel Estimator is proposed and it is successfully implemented in signal processing simulation system. Then the simulated performances are compared with the conventional Time domain linear interpolated estimator. According to the simulation results, QPSK modulation can be used even under severe channel condition such as 500 km/h, 2 path reverse Doppler Shift condition, although QPSK modulation can be used less than 200 km/h with conventional Channel estimator.

• Journal of Information Processing Systems
• /
• 제13권1호
• /
• pp.141-151
• /
• 2017

Effective identification of wireless channel in different scenarios or regions can solve the problems of multipath interference in process of wireless communication. In this paper, different characteristics of wireless channel are extracted based on the arrival time and received signal strength, such as the number of multipath, time delay and delay spread, to establish the feature vector set of wireless channel which is used to train backpropagation (BP) neural network to identify different wireless channels. Experimental results show that the proposed algorithm can accurately identify different wireless channels, and the accuracy can reach 97.59%.

• 한국전자통신학회논문지
• /
• 제5권4호
• /
• pp.357-362
• /
• 2010

본 논문에서는 음성 및 잡음 구간을 검출하기 위하여 신경회로망에 의한 음성 및 잡음 인식시스템을 제안한다. 제안하는 신경회로망은 오차역전파알고리즘에 의하여 학습되는 네트워크이다. 먼저, 고속 푸리에변환에 의한 전력스펙트럼 및 선형예측계수가 각 프레임에서 신경회로망의 입력으로 사용되어 네트워크가 학습된다. 따라서 제안된 신경회로망은 잡음이 중첩되지 않은 음성 및 잡음을 사용하여 학습된다. 제안한 인식시스템의 성능은 다양한 음성 및 백색, 프린터, 도로, 자동차 잡음 들을 사용하여 인식율에 의하여 평가된다. 본 실험에서는 신경회로망의 학습 데이터 및 평가 데이터가 다를 경우에도 이러한 음성 및 잡음에 대하여 92% 이상의 인식율을 구할 수 있었다.

• International journal of advanced smart convergence
• /
• 제9권2호
• /
• pp.58-67
• /
• 2020

In this paper, we propose a data-driven-based beam selection scheme for massive multiple-input and multiple-output (MIMO) systems in ultra-dense networks (UDN), which is capable of addressing the problem of high computational cost of conventional coordinated beamforming approaches. We consider highly dense small-cell scenarios with more small cells than mobile stations, in the millimetre-wave band. The analog beam selection for hybrid beamforming is a key issue in realizing millimetre-wave UDN MIMO systems. To reduce the computation complexity for the analog beam selection, in this paper, two deep neural network models are used. The channel samples, channel gains, and radio frequency beamforming vectors between the access points and mobile stations are collected at the central/cloud unit that is connected to all the small-cell access points, and are used to train the networks. The proposed machine-learning-based scheme provides an approach for the effective implementation of massive MIMO system in UDN environment.

• 한국정보통신학회:학술대회논문집
• /
• 한국정보통신학회 2014년도 춘계학술대회
• /
• pp.893-895
• /
• 2014

무선통신기반 열차제어시스템의 국내 도입을 앞두고 철도전용 주파수 할당에 대한 사항이 활발히 논의되고 있다. 공공자원으로서 큰 가치를 지닌 주파수를 적절한 용도에 알맞은 양을 할당하는 것은 다른 분야와의 형평성과 한정된 자원의 유용한 사용 측면에 있어서도 중요한 문제이다. 본 논문에서는 무선통신기반 열차제어시스템에서 기존에 논의되던 서비스들을 LTE 기반의 철도통신망에 적용했을 때 안정적인 운용이 가능한지에 대하여 시뮬레이션을 통하여 알아보았다.

• IEIE Transactions on Smart Processing and Computing
• /
• 제3권4호
• /
• pp.212-220
• /
• 2014

The current cellular communications are optimized for low mobility users, meaning that their performance is degraded at high speed. Therefore, passengers in a high-speed train experience very poor radio link quality due to the significantly large number of simultaneous handovers. In addition, wireless data traffic is expanding exponentially in trains, subways and buses due to the widespread use of smartphones and mobile devices. To solve the inherent problem of cellular communication networks and meet the growing traffic demand, this paper proposes the mobile hotspot network of a millimeter-wave communication system as a mobile wireless backhaul. This paper describes the physical layer design of uplink and downlink in the proposed system, and the performances of uplink and downlink are evaluated under Rician fading channel conditions. The implemented baseband prototype of the proposed millimeter-wave communication modem is presented. This system can provide a Gbps data rate service in high-speed trains carrying hundreds of wireless Internet users.

• 대한전기학회:학술대회논문집
• /
• 대한전기학회 2005년도 제36회 하계학술대회 논문집 B
• /
• pp.1562-1564
• /
• 2005

The study for ubiquitous computing infra is proceeding actively, it make possible to use service and access network anywhere, anytime because of wire/wireless communication technology and progress of hardware. Domestically, study for the network mobility support technology which is the key technology for future ubiquitous computing realization have progressed, but that is insufficient. Especially, there is no study for independent mobility support study about railway wireless network. So, this study propose network mobility management technology for mobile network infra in railway and proper network model in train.

• 한국산학기술학회논문지
• /
• 제17권11호
• /
• pp.202-207
• /
• 2016

장대 터널 및 급곡선 구간과 같은 시계 확보가 어려운 지역에서 위험 및 안전 정보를 함께 공유하기 위해서는 기관사, 열차 관제사 및 현장 작업자 사이에 보조적인 실시간 열차 정보 통신시스템의 개발이 필요하다. 이에 적합한 열차 정보 통신시스템을 개발하기 위하여 본 연구에서는 작은 노이즈와 신호 감쇄를 동시에 갖도록 하는 레일 전력선 통신을 제안하였다. 이에 대한 레일 전력선 통신의 기초 연구 일환으로서, 본 통신기법의 전송 특성을 향상할 수 있는 레일에 대한 임피던스 측정 및 정합특성 평가가 중요하다. 본 연구에서는 자체 제작한 임피던스 정합 트랜스포머와 네트워크 분석기를 이용하여 레일에 대한 반사 및 전파 특성을 평가하였다. 대향 접속방식의 임피던스 반사특성 시험결과로부터 레일 전력선 통신에 적합한 정합용 트랜스포머의 정합비율은 1:3.28임을 알 수 있었다. 또한, 제작한 임피던스 정합용 트랜스포머는 레일 전력선통신에서 투과 신호 전송을 향상할 수 있음을 확인하였다.

• 대한전기학회논문지:전기기기및에너지변환시스템부문B
• /
• 제55권9호
• /
• pp.485-490
• /
• 2006

According to the computerization of railway signaling systems, the interface link between the signaling systems has been replaced by the digital communication channel. At the same time, the importance of the communication link is more pronounced than before. In this paper, new Network-based protocol between railway signaling and SCADA (Supervisory Control and Data Acquisition system) has designed and the overview of designed protocol is briefly represented. And also this paper addresses analysis of newly designed train control systems. Fame error rates of the data transmissions are calculated and compared for the two cases that the CTC (Centralized Traffic Control)/SCADA has an extra data transmission error control (CRC16) besides the inherent error control of the Ethernet and that the CTC/SCADA has no extra data transmission error control. With simulation results it has been verified that the additional error control code contributes to lowering the frame error rate. It will be expected to increase the safety, reliability and efficiency of maintenance of the signaling systems by using the designed protocol for railway signaling system.