• Journal of IKEEE
• /
• v.22 no.2
• /
• pp.402-407
• /
• 2018

Link selection strategy has been an important technical issues of relay network. In this paper, we introduce a link selection scheme in the bidirectional, buffer-aided relay network. Three kinds of information such as the states of the queue at the relay buffer, the qualities of the links, and the states of the queues at the user buffer are considered. Throughput and delay performance is evaluated under three cases with different available information.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.17 no.1
• /
• pp.136-142
• /
• 2007

This paper proposes a new approach on nonlinear maneuvering target tracking. In this paper, proposed algorithm is the Kalman filter based on the adaptive interactive multiple model using the concept of predicted impact point and utilize modified Kalman filter regarding the error between measurement position and predicted impact point. The unknown target acceleration is regarded as an additional process noise to the target model, and each sub-model is characterized in accordance with the valiance of the overall process noise which is obtained on the basis of each acceleration interval. To compensate the decreasing performance of Kalman filter in nonlinear maneuver, we construct optional algorithm to utilize proposed method or Kalman filter selectively. To effectively estimate the acceleration during the target maneuvering, the rapid increase of the noise scale is recognized as the acceleration to be used in maneuvering target's movement equation. And a few examples are presented to show suggested algorithm's executional potential.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.19 no.7
• /
• pp.24-35
• /
• 2018

In wireless communications environments, a link adaptation technique that selects the proper rate from among several transmission rates is adopted to cope with variations in channel status. In block coded modulation, the frame time and/or the block length can be adjusted to the channel status. A smaller frame time can cause inter-frame interference (IFI), however, a larger frame time can reduce the data rate. Therefore, frame time is the design factor decided by a tradeoff between performance and the data rate. This paper presents a method to determine the frame time based on the processing gain for the channel model, CM1~CM4, recommended by IEEE 802.15a. Also, a link adaptation technique for block coded modulation is proposed for efficient communications by varying the frame time and the block length. Through simulation, link adaptation can be carried out with a step size of 2~5 nsec in a frame time range of 14~ 50 nsec for channel models CM1~CM4.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.46 no.6
• /
• pp.163-174
• /
• 2009

Applying the maximally polyphase decomposition and noble identity to the adaptive filter in subband structure, the conventional fullband affine projection algorithm is translated to the subband affine projection (SAP) algorithm. The Maximally polyphase decomposed SAP (MPDSAP) algorithm is a special version of the SAP algorithm, and its adaptive sub-filters have unity projection dimension. The weight updating formular of the MPDSAP is similar to that of the NLMS algorithm, so it may be more proper algorithm than other AP-type algorithms for many practical applications. This paper presents a new statistical analysis of the MPDSAP algorithm. The analytical model is derived for autoregressive (AR) inputs and the nonunity adaptive gain in the subband structure with the orthonormal analysis filters (OAF), The pre-whitening by the OAF allows the derivation of a simple-analytical model for the MPDSAP with the AR inputs and the nonunity adaptive gain.

• The KIPS Transactions:PartD
• /
• v.15D no.1
• /
• pp.73-86
• /
• 2008

Because of its capability of automatic identification of objects, RFID(Radio Frequency Identification) technologies have extended their application areas to logistics, healthcare, and food management system. Load balancing is a basic technique for improving scalability of systems by moving loads of overloaded middlewares to under loaded ones. Adaptive load balancing has been known to be effective for distributed systems of a large load variance under unpredictable situations. There are needs for applying load balancing to RFID middlewares because they must efficiently treat vast numbers of RFID tags which are collected from multiple RFID readers. Because there can be a large amount of variance in loads of RFID middlewares which are difficult to predict, it is desirable to consider adaptive load balancing approach for RFID middlewares, which can dynamically choose a proper load balancing strategy depending on the current load. This paper proposes an adaptive load balancing approach for RFID middlewares and presents its design and implementation. First we decide a performance model by a experiment with a real RFID middleware. Then, a set of proper load balancing strategies for high/medium/low system loads is determined from a simulation of various load balancing strategies based on the performance model.

• Proceedings of the Korean Institute of Intelligent Systems Conference
• /
• 2004.10a
• /
• pp.309-312
• /
• 2004

본 논문에서는 인간과 로봇의 상호작용을 위해 감정에 기반한 감정 처리 모델을 설계하였다. 감정 재현 기술은 사용자에게 친근감을 주기 위해 로봇 시스템이 제스처, 표정을 통하여 사람이나 동물의 감성과 동작을 표현하는 분야이다. 로봇이 감정을 표현하는 문제에는 많은 심리학적, 해부학적, 공학적 문제가 관련된다. 여러가지 애매모호한 상황임에 불구하고 심리학자인 Ekman과 Friesen에 의해 사람의 여섯 가지 기본 표정이 놀람, 공포, 혐오, 행복감, 두려움, 슬픔은 문화에 영향을 받지 않고 공통적으로 인식되는 보편성을 가지고 있는 것으로 연구됐다. 사람의 행동에 대한 로봇의 반응이 학습되어 감정모델이 결정되고, 그 결과가 행동결정에 영향을 주어 로봇의 행동에 반영되도록 하였다. 본 논문에서는 인간과 로봇과의 상호작용을 통해 정보를 축적하고 인간의 반응에 적응해나 갈 수 있는 감정 처리 모델을 제안한다.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2011.11a
• /
• pp.314-317
• /
• 2011

본 논문은 상향식 현저함 모델을 이용하여 입력 영상으로부터 시각적 주의를 갖는 영역들을 자동으로 검출하는 방법을 제안한다. 제안한 방법에서는 인간의 시각 시스템과 같이 사전 지식 없이 시각정보의 공간적인 분포에 근거하여 장면을 해석하는 상향식 현저함 모델 방법을 입력 영상에 적용하여 관심 물체 영역을 검출하는 연구이다. 상향식 현저함 방법은 Treisman의 세부특징이론 연구에서 제시한 바와 같이 시각적 주의를 갖는 영역은 시각정보의 현격한 대비차이를 가지는 영역으로 집중되어 배경에서 관심영역을 구분할 수 있다. 입력 영상에서 현저함 모델을 통해 3차원 현저함 맵을 생성한다. 그리고 생성된 현저함 맵으로부터 실제 관심영역들을 검출하기 위해 제안한 방법에서는 적응적 임계치 방법을 적용하여 관심영역을 검출한다. 제안한 방법을 관심영역 분할에 적용한 결과, 영역 분할 정확도 및 정밀도가 약 88%와 89%로 제시되어 관심 영상분할 시스템에 적용이 가능함을 알 수 있다.

• The Journal of Bigdata
• /
• v.9 no.1
• /
• pp.101-113
• /
• 2024

Federated learning has garnered attention as an efficient method for training machine learning models in a distributed environment while maintaining data privacy and security. This study proposes a novel FedRFBagging algorithm to optimize the performance of random forest models in such federated learning environments. By dynamically adjusting the trees of local random forest models based on client-specific data characteristics, the proposed approach reduces communication costs and achieves high prediction accuracy even in environments with numerous clients. This method adapts to various data conditions, significantly enhancing model stability and training speed. While random forest models consist of multiple decision trees, transmitting all trees to the server in a federated learning environment results in exponentially increasing communication overhead, making their use impractical. Additionally, differences in data distribution among clients can lead to quality imbalances in the trees. To address this, the FedRFBagging algorithm selects only the highest-performing trees from each client for transmission to the server, which then reselects trees based on impurity values to construct the optimal global model. This reduces communication overhead and maintains high prediction performance across diverse data distributions. Although the global model reflects data from various clients, the data characteristics of each client may differ. To compensate for this, clients further train additional trees on the global model to perform local optimizations tailored to their data. This improves the overall model's prediction accuracy and adapts to changing data distributions. Our study demonstrates that the FedRFBagging algorithm effectively addresses the communication cost and performance issues associated with random forest models in federated learning environments, suggesting its applicability in such settings.

• The Journal of the Acoustical Society of Korea
• /
• v.30 no.6
• /
• pp.315-323
• /
• 2011

This paper proposes two methods of estimating prior distribution to improve the performance of rapid speaker adaptation based on maximum a posteriori linear regression (MAPLR). In general, prior distribution of the transformation matrix used in MAPLR adaptation is estimated from all of the training speakers who are employed to construct the speaker-independent model, and it is applied identically to all new speakers. In this paper, we propose a method in which prior distribution is estimated from a group of reference speakers, selected using adaptation data, so that the acoustic characteristics of the selected reference speakers may be similar to that of the new speaker. Additionally, in MAPLR adaptation with block-diagonal transformation matrix, we propose a method in which the mean matrix and covariance matrix of prior distribution are estimated from two groups of transformation matrices obtained from the same training speakers, respectively. To evaluate the performance of the proposed methods, we examine word accuracy according to the number of adaptation words in the isolated word recognition task. Experimental results show that, for very limited adaptation data, statistically significant performance improvement is obtained in comparison with the conventional MAPLR adaptation.

• Journal of the Korea Concrete Institute
• /
• v.14 no.6
• /
• pp.1022-1031
• /
• 2002

This paper presents efficient models for bridge structures using CART-ANFIS (classification and regression tree-adaptive neuro fuzzy inference system). A fuzzy decision tree partitions the input space of a data set into mutually exclusive regions, each region is assigned a label, a value, or an action to characterize its data points. Fuzzy decision trees used for classification problems are often called fuzzy classification trees, and each terminal node contains a label that indicates the predicted class of a given feature vector. In the same vein, decision trees used for regression problems are often called fuzzy regression trees, and the terminal node labels may be constants or equations that specify the predicted output value of a given input vector. Note that CART can select relevant inputs and do tree partitioning of the input space, while ANFIS refines the regression and makes it continuous and smooth everywhere. Thus it can be seen that CART and ANFIS are complementary and their combination constitutes a solid approach to fuzzy modeling.