• Annual Conference on Human and Language Technology
• /
• 2019.10a
• /
• pp.165-169
• /
• 2019

자연어 이해 모델은 대화 시스템의 핵심적인 구성 요소로서 자연어 문장에 대해 그 의도와 정보를 파악하여 의도(intent)와 슬롯(slot)의 형태로 분석하는 모델이다. 최근 연구에서 의도와 슬롯의 추정을 단일 합동 모델(joint model)을 이용하여 합동 학습(joint training)을 하는 연구들이 진행되고 있다. 합동 모델을 이용한 합동 학습은 의도와 슬롯의 추정 정보가 모델 내에서 암시적으로 교류 되도록 하여 의도와 슬롯 추정 성능이 향상된다. 본 논문에서는 기존 합동 모델이 암시적으로 추정 정보를 교류하는 데서 더 나아가 모델 내의 의도와 슬롯 추정 정보를 명시적으로 교류하도록 모델링하여 의도와 슬롯 추정 성능을 높일 수 있는 교차 게이트 메커니즘(Cross Gated Mechanism)을 제안한다.

• Proceedings of the IEEK Conference
• /
• 2000.11a
• /
• pp.169-172
• /
• 2000

본 논문에서는 다중 경로상의 DS-CDMA 시스템을 위해 우리가 이미 제안한 Rake 수신기와 다이버시티 결합 구조를 이용하는 CMA 기반의 적응 간섭 제거 기법에서 실제 채널 추정 기법을 사용할 때 부정확한 채널추정이 성능에 미치는 영향을 검증하였으며, sliding widow averaging 형태의 단순한 채널 추정 기법을 이용할 경우 slow 페이딩 및 fast 페이딩 환경에서의 모의실험 결과 이상적인 채널 추정을 가정한 경우와 유사한 우수한 성능을 보임을 확인하였다.

• Proceedings of the IEEK Conference
• /
• 2001.09a
• /
• pp.829-832
• /
• 2001

본 논문에서는 2차 통계치를 이용하여 패널추징 및 등화 기법을 제안하였다. 기존의 채널 추정 알고리듬은 잡음이 없는 환경에서 LS방법을 이용하기 때문에 잡음이 강한 패널에서는 원하는 성능을 얻을 수 없는 단점이 있다. 수신신호의 상관행렬의 최소 고유값에 대응하는 고유벡터는 채널의 임펄스 응답에 관한 정보를 포함하고 있다. 이러한 고유 벡터를 매시간마다 갱신시키면서 구하는 적응 알고리듬을 제안하고 이를 이용하여 블라인드 채널 추정 및 등화기 파라미터를 추정하였다. 제안한 알고리듬은 잡음에 강인한 특성을 보일 뿐 아니라 기존의 알고리듬들 보다 우수한 채널 추정 및 등화 성능을 모의 실험을 통하여 검증하였다.

• The Journal of Korean Institute of Electromagnetic Engineering and Science
• /
• v.22 no.3
• /
• pp.282-291
• /
• 2011

Adaptive modulation(AM) is an important technique to increase the system efficiency, in which transmitter selects the most suitable modulation mode adaptively according to channel state in the temporary and spatially varying communication environment. Fixed modulation on channels with varying signal-to-noise ratio(SNR) is that the bit-errorrate(BER) probability performance is changing with the channel quality. An adaptive modulation scheme can be designed to have a BER which is constant for all channel SNRs. The correct as well as fast and simple SNR estimation is required essentially for this adaptive modulation. In order to operate adaptive modulation system effectively, in this paper, we analyze the effect of SNR estimation performance to it through the average BER and data throughput. Applying SNR estimation based on auto-correlation of decision feedback signal and others to adaptive modulation system, we also confirm performance degradation or improvement of its which is decided by SNR estimation error at each transition point of modulation level. Since SNR estimation based on auto-correlation of decision feedback signal shows stable estimation performance for various quadrature amplitude modulation(QAM) comparatively, this can be reduced degradation than others at each transition point of modulation level.

• Journal of Advanced Navigation Technology
• /
• v.10 no.4
• /
• pp.385-393
• /
• 2006

In this paper, we proposed a location estimation algorithm considering an extension of searching area in 2.45GHz band RTLS and analyzed its performance in terms of an average estimation error distance. The extendable searching area was assumed to be square of $300m{\times}300m$ and 2 dimensions. The arrangement shape of available readers was considered circle, rectangle, and shrinkage rectangle for extendable searching area. Also, we assumed that propagation path was LOS (Line-Of-Sight) environment, and analyzed the estimation error performance as a function of the number of received sub-blink considering an arrangement shape of available readers in searching area. From the results, compared with rectangle shape, circle shape showed the higher estimation accuracy. Also, we confirmed that the proposed location estimation algorithm provided high estimation accuracy in the shrinkage rectangle shape that was suitable for extension of searching area.

• Journal of Advanced Navigation Technology
• /
• v.18 no.4
• /
• pp.260-267
• /
• 2014

For the purpose of improving the accuracy of Wide Area Differential GNSS (WA-DGNSS), estimation performance of ionospheric delay error which has a great impact on GPS error sources should be enhanced. This paper applied multi-constellation GNSS which represents GPS in USA, GLONASS in Russia, and Galileo in Europe to WA-DGNSS algorithm in order to improve performance of ionospheric delay estimation. Furthermore, we conducted simulation to analyze ionospheric delay estimation performance in Korean region by increasing the number of reference stations. Consequently, using multi-constellation GNSS to improve performance of ionospheric delay estimation is more effective than increasing the number of reference stations in spite of similar number of measurements which are in use for estimation. We expect this result can contribute to improvement for ionospheric delay estimation performance of single-frequency SBAS (Satellite Based Augmentation System) user.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2021.06a
• /
• pp.180-182
• /
• 2021

본 논문은 눈 랜드마크 위치 검출과 시선 방향 벡터 추정이 하나의 딥러닝 네트워크로 통합된 시선 추정 네트워크를 제안한다. 제안하는 네트워크는 Stacked Hourglass Network[1]를 백본(Backbone) 구조로 이용하며, 크게 랜드마크 검출기, 특징 맵 추출기, 시선 방향 추정기라는 세 개의 부분으로 구성되어 있다. 랜드마크 검출기에서는 눈 랜드마크 50개 포인트의 좌표를 추정하며, 특징 맵 추출기에서는 시선 방향 추정을 위한 눈 이미지의 특징 맵을 생성한다. 그리고 시선 방향 추정기에서는 각 출력 결과를 조합하고 이를 통해 최종 시선 방향 벡터를 추정한다. 제안하는 네트워크는 UnityEyes[2] 데이터셋을 통해 생성된 가상의 합성 눈 이미지와 랜드마크 좌표 데이터를 이용하여 학습하였으며, 성능 평가는 실제 사람의 눈 이미지로 구성된 MPIIGaze[3] 데이터 셋을 이용하였다. 실험을 통해 시선 추정 오차는 0.0396 MSE(Mean Square Error)의 성능을 보였으며, 네트워크의 추정 속도는 42 FPS(Frame Per Second)를 나타내었다.

• Proceedings of the Korean Society for Noise and Vibration Engineering Conference
• /
• 2008.04a
• /
• pp.909-915
• /
• 2008

Researches for time delay estimation had been studied broadly. However studies about factors for time delay estimation are insufficient, especially in case of real environment application. In 1997, Brandstein and Siverman announced that performance of time delay estimation deteriorates as reverberant time of room increases. Even though reverberant time of room is same, performance of estimation is different as the specific part of signals. In order to know that reason, we studied and analyzed the factors for time delay estimation using speech signal and room impulse response. In result, we can know that performance of time delay estimation is changed by different R/D ratio and signal characteristics in spite of same reverberant time.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.9 no.2
• /
• pp.395-400
• /
• 2008

This paper relates to the performance analysis of DoA estimation algorithm in 2-dimensional circular array antenna for the receiving of GPS signal which is used for the performance improvement by elimination of jammer signal. By performing the spatial filtering after the DoA estimation in array antenna, the quality of receiving signal can improve by the nulling of jammer signal from the undesired direction and the forming of beam from the desired direction. In this paper, the MUSIC and MinNorm algorithm used for DoA estimation were applied after fixing the angle and power of jammer signal in 4 element and 7 element circular array antenna. In order to performance analysis, the estimation result and estimation error were computed by computer simulation. As a result, the MUSIC and MinNorm were fairly good in azimuth and elevation angle estimation of DoA in case of good signal to noise ratio and the MUSIC has better performance compared to MinNorm in case of poor signal to noise ratio.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.37C no.12
• /
• pp.1245-1255
• /
• 2012

There are various anti-jamming algorithms for the GNSS system which is vulnerable to jamming, and the methods using array antenna show the best performance. Among them, the DOA estimation algorithms to identify the location of the jammer is very important. However, in case of the rotorcraft, the wireless channel which amplitude and phase changes with time is generated by the rotation of the rotor and it affects the performance of existing anti-jamming algorithms. In this paper, we modeled the effect of the rotor in four scenarios according to the correlation of antennas and assured that the performance of DOA estimation algorithms are degraded and saturated regardless of JNR due to the rotor effect. When we use the averaging method to solve this problem, the performance is improved as increasing samples for estimating. And in case of using moving average method with averaging, it shows similar performance. In addition, it reduces the required memory and moderates the variation of DOA estimation.