• 대한전자공학회논문지SP
• /
• 제39권2호
• /
• pp.116-127
• /
• 2002

본 논문에서는 CDHMM 음성인식기의 인식성능을 향상시키기 위해 상태 당 관측밀도함수 수 변화에 의한 화자적응 알고리듬을 제안하였다. 제안한 방법은 CDHMM의 각 상태마다 관측 확률밀도함수의 가지 수가 두 개 이상이 릴 수도 있게 하여 발음특성의 다양성을 반영할 수 있게 하였다. 가지 수는 각 상태에 속하는 적응음성의 프레임 수에 따라 정하는 방법과 특징벡터 행렬식에 따라 정하는 방법으로 하였다 이두 방법중의 어느 하나로 관측 확률밀도함수의 가지가 결정되면, 세분화된 각 가지로부터 MAP 파라미터를 추출함으로써 정밀한 화자적응모델의 파라미터를 구할 수 있었다. 아울러 적응음성을 상태분할 할 때 기존의 화자독립모델을 사전정보로 이용함으로써 ML 추정시의 초기 상태분할 오류의 영향을 줄여 기존 상태분 할 방법의 단점을 개선하였다 그리고 상태지속분포를 화자에 적응시킴으로써 화자 고유의 발음속도와 발음 패턴 등의 음성특성을 흡수하도록 하였다. 제안한 방법들의 타당성을 확인하기 위한 실험에서 제안한 방법이 기존 방법에 비해 높은 인식률을 얻음을 확인하였다.

• 한국정보과학회논문지:정보통신
• /
• 제30권6호
• /
• pp.755-763
• /
• 2003

전파(propagation) 오류가 빈번한 무선 이동 네트워크에서는 전송 성능을 향상하기 위해 FEC(Forward Error Correction)알고리즘을 채택한다. 그러나 정적인 FEC방식은 연속적으로 변화하는 전파 오류율에 알맞은 정정 코드(check code)를 적용하지 못해 성능이 저하된다. 본 논문에서는 변화하는 무선 채널의 전파 오류율에 따라 FEC의 정정도를 알맞게 결정하는 링크 계층용 적응적 FEC기법인 FECA(FEC-level Adaptation)를 제안한다. FECA는 오류율이 높고, 오류율이 천천히 변화하는 무선 환경에 알맞은 알고리즘이다. 일례로 전파 간섭이 있는 환경에서 센서(sensor) 네트워크는 평균 오류율이 $10^{-6}$이상이며 오류율이 평균 수백 밀리초 이상 지속되는 것으로 관찰되었다. FECA는 분석적인 무선채널 시뮬레이션과 패킷 트레이스 기반(trace-driven) 시뮬레이션에서 정적 FEC 알고리즘에 비해 최대 15%이상 성능을 향상하였다.

• ETRI Journal
• /
• 제27권3호
• /
• pp.257-266
• /
• 2005

Many multi-modulus blind equalization algorithms (MMA) have been presented in the past to overcome the undesirable high misadjustment exhibited by the well-known constant modulus algorithm. Some of these MMA schemes, specifically tailored for quadrature amplitude modulation (QAM) constellations, have also been proved to fix the phase offset error without needing any rotator at the end of the equalizer stage. In this paper, a new multi-modulus algorithm is presented for QAM signals. The contribution lies in the technique to incorporate the sliced symbols (outcomes of decision device) in the multi-modulus-based weight adaptation process. The convergence characteristics of the proposed sliced multi-modulus algorithm (S-MMA) is demonstrated by way of simulations, and it is shown that it gives better steady-state performance in terms of residual inter-symbol interference and symbol-error rate. It has also been shown that the proposed algorithm exhibits lesser steady-state misadjustment compared to the best reported MMA.

• 대한전기학회:학술대회논문집
• /
• 대한전기학회 2002년도 하계학술대회 논문집 D
• /
• pp.2824-2826
• /
• 2002

Immune system is an evolutionary biological system to protect innumerable foreign materials such as virus, germ cell, and etcetera. Immune algorithm is the modeling of this systems response that has adaptation and reliability when disturbance occur. In this paper, immune algorithm is proposed to control four wheels steering AGV(Automated Guided Vehicle) in container yard. The adaptive immune system is applied to the PID controller. For design the PID controller using immune algorithm, we tune PID parameters by off-line manner, in order to avoid the damage from abrupt control force. Repeatedly, the PID parameters are adjusted to be accurate by on-line fine tuner of immune algorithm. And then the computer simulation result from the viewpoint of yaw rate and lateral displacement are analyzed and compared with result of conventional PID controller.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제7권6호
• /
• pp.1463-1479
• /
• 2013

In this paper, we propose an unequal protection physical coding sub-layer (PCS) forward error correction (FEC) scheme for efficient and high-quality transmission of video data over optical access networks. Through identifying and resolving the unequal importance of different video frames and passing this importance information from MAC-layer to PCS, FEC scheme of PCS can be adaptive to application-layer data. Meanwhile, we jointly consider the different channel situations of optical network unit (ONU) and improve the efficiency of FEC redundancy by channel adaptation. We develop a theoretical algorithm and a hardware method to achieve efficient FEC assignment for the proposed unequal protection scheme. The theoretical FEC assignment algorithm is to obtain the optimal FEC redundancy allocation vector that results in the optimum performance index, namely frame error rate, based on the identified differential importance and channel situations. The hardware method aims at providing a realistic technical path with negligible hardware cost increment compared with the traditional FEC scheme. From the simulation results, the proposed Channel and Application-layer data Adaptation Unequal Protection (CAAUP) FEC scheme along with the FEC ratio assignment algorithm and the hardware method illustrates the ability of efficient and high-quality transmission of video data against the random errors in the channel of optical access networks.

• 대한전자공학회논문지TE
• /
• 제39권2호
• /
• pp.17-28
• /
• 2002

최근 신경회로망 구조 또는 퍼지논리를 이용한 자력등화 즉, 비선형 등화에 대한 연구가 활발하게 진행되고 있다. 본 논문에서는 계산의 복잡성과 등화 성능 사이에 더 나은 trade-off를 찾기 위해 스텝 크기(step size)를 자동적으로 조절할 수 있는 즉, 에러 오차 값이 크면 계수갱신율을 크게 하고 에러 오차 값이 작으면 계수갱신율을 작게 하여 빠른 수렴비와 낮은 초과 MSE를 갖는 TS(Tagaki-Sugeno) 퍼지 모델과 ISI에 강하고 위상변화에 둔감한 CMA(Constant Modulus Algorithm)를 접목시킨 새로운 FSG (Fuzzy Stochastic Gradient)/CMA 알고리즘을 제안하였다. 제안된 FSG/CMA 알고리즘을 사용한 채널 적응 등화기를 설계하여 모의실험한 결과 기존 알고리즘 보다 3.5dB의 SNR이 개선됨을 확인하였다.

• 한국정보과학회:학술대회논문집
• /
• 한국정보과학회 2006년도 한국컴퓨터종합학술대회 논문집 Vol.33 No.1 (D)
• /
• pp.136-138
• /
• 2006

The high-rate Wireless Personal Area Networks (WPANs) which IEEE 802.15.3 standard support, foster the Digital Home Network (DHN) scenarios with high rate multimedia data transmission. Actually, there are a few routing protocols for ad-hoc networks which considered the terminal location information and routing metric to reduce the energy consumption and optimize the routing path in mobile system. Based on other routing protocols, this paper presents the reliable location-based routing algorithm which is an adaptation to these networks.

• 대한음성학회지:말소리
• /
• 제56호
• /
• pp.147-158
• /
• 2005

In this paper, we are focused to develop the simplified speaker verification algorithm without background speaker models, which will be adopted in the portable speaker verification system equipped in portable terminals such as mobile phone and PMP. According to the tolerance interval analysis, the population of someone's speaker model can be represented by a suitable number of selected independent samples of speaker model. So we can make the representative speaker model and threshold under the specified confidence level and coverage. Using proposed algorithm with the number of samples is 40, the experiments show that the false rejection rate is $3.0\%$ and the false acceptance rate $4.3\%$, worth comparing to conventional method's results, $5.4\%\;and\;5.5\%$, respectively. Next step of research will be on the suitable adaptation methods to overcome speech variation problems due to aging effect and operating environments.

• 전자공학회논문지B
• /
• 제33B권2호
• /
• pp.187-196
• /
• 1996

This paper introduces the concept of fixed weights and proposes an algorithm for classification by adding this concept to vector space separation method in LVQ. The proposed algorithm is based on competitive learning. It uses fixed weightsfor generality and fast adaptation efficient radius for new weight creation, and L1 distance for fast calcualtion. It can be applied to many fields requiring adaptive learning with the support of generality, real-tiem processing and sufficient training effect using smaller data set. Recognition rate of over 98% for the train set and 94% for the test set was obtained by applying the suggested algorithm to on-line handwritten recognition.

• Journal of Communications and Networks
• /
• 제14권6호
• /
• pp.619-628
• /
• 2012

The ZigBee network has been considered to monitor electricity usage of home appliances in the smart grid network. ZigBee, however, may suffer from a coexistence problem with wireless local area network (WLAN). In this paper, to resolve the coexistence problem between ZigBee network and WLAN, we propose a new protocol constructing a cognitive smart grid network for supporting monitoring of home appliances. In the proposed protocol, home appliances first estimates the transmission timing and channel information of WLAN by reading request to send/clear to send (RTS/CTS) frames of WLAN. Next, based on the estimated information, home appliances transmit a data at the same time as WLAN transmission. To manage the interference between WLAN and smart grid network, we propose a cognitive beamforming algorithm. The beamforming algorithm is designed to guaranteeing zero interference to WLAN while satisfying a required rate for smart metering. We also propose an energy efficient rate adaptation algorithm. By slowing down the transmission rate while satisfying an imperceptible impact of quality of service (QoS) of the receiver, the home appliance can significantly save transmit power. Numerical results show that the proposed multiple antenna technique provides reliable communications for smart metering with reduced power comparing to the simple transmission technique.