• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.29 no.1A
• /
• pp.84-92
• /
• 2004

Eigenstructure-based adaptive beamfoming has advantages of fast convergence and the insentivity to errors in the arrival angle of the desired signal. Eigen-decomposing the sample matrix to extract a basis for the Sl (signal plus interference) subspace, however, is very computationally expensive. In this paper, we present a simple subspace based beamforming which utilizes off-diagonal elements of the sample matrix to estimate the Sl subspace. The outputs of overlapped subarrays are combined to produce the final adaptive output, which improves SINR (signal-to-interference-plus-noise ratio) comapred to exploiting a single subarray. The proposed adaptive beamformer, which employs an efficient angle estimation is very roubust to errors in both the arrival angles and the number of the incident signals, while the eigenstructure-based beamforer suffers from severe performance degradation.

• The Journal of the Acoustical Society of Korea
• /
• v.39 no.5
• /
• pp.490-496
• /
• 2020

We present a method to improve the performance of the multi-channel Wiener filter in noisy environment. To build subspace-based multi-channel Wiener filter, in the case of single target source, the target speech component can be effectively estimated in the principal subspace of speech correlation matrix. The speech correlation matrix can be estimated by subtracting noise correlation matrix from signal correlation matrix based on the assumption that the cross-correlation between speech and interfering noise is negligible compared with speech correlation. However, this assumption is not valid in the presence of strong interfering noise and significant error can be induced in the principal subspace accordingly. In this paper, we propose to adjust the principal subspace vector using speech presence probability and the steering vector for the desired speech source. The multi-channel speech presence probability is derived in the principal subspace and applied to adjust the principal subspace vector. Simulation results show that the proposed method improves the performance of multi-channel Wiener filter in noisy environment.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.48 no.4
• /
• pp.153-159
• /
• 2011

The MUSIC (MUltiple SIgnal Classification) method estimates the directions of arrival (DOAs) of the signals impinging on a sensor array based on the fact that the noise subspace is orthogonal to the signal subspace. In the conventional MUSIC, an estimate of the basis for the noise subspace is obtained by eigendecomposing the sample matrix, which is computationally expensive. In this paper, we present a simple DOA estimation method which finds an estimate of the signal subspace basis directly from the columns of the sample matrix from which the noise power components are removed. DOA estimates are obtained by searching for minimum points of a cost function which is defined using the estimated signal subspace basis. The minimum points are efficiently found through the Brent method which employs parabolic interpolation. Simulation shows that the simple estimation method virtually has the same performance as the complex conventional method based on the eigendecomposition.

• 제어로봇시스템학회:학술대회논문집
• /
• 2000.10a
• /
• pp.525-525
• /
• 2000

In this paper, we present recursive algorithms for state space model identification using subspace extraction via Schur complement. It is shown that an estimate of the extended observability matrix can be obtained by subspace extraction via Schur complement. A relationship between the least squares residual and the Schur complement matrix obtained from input-output data is shown, and the recursive algorithms for the subspace-based state-space model identification (4SID) methods are developed. We also proposed the above algorithm for an instrumental variable (IV) based 4SID method. Finally, a numerical example of the application of the algorithms is illustrated.

• Proceedings of the Computational Structural Engineering Institute Conference
• /
• 1999.10a
• /
• pp.359-369
• /
• 1999

The paper describes the study of concurrent subspace optimization(CSSO) for coupled multidisciplinary design optimization (MDO) techniques in mechanical systems. This method is a solution to large scale coupled multidisciplinary system, wherein the original problem is decomposed into a set of smaller, more tractable subproblems. Key elements in CSSO are consisted of global sensitivity equation(GSE), subspace optimization (SSO), optimum sensitivity analysis(OSA), and coordination optimization problem(COP) so as to inquiry valanced design solutions finally, Automatic differentiation has an ability to provide a robust sensitivity solution, and have shown the numerical numerical effectiveness over finite difference schemes wherein the perturbed step size in design variable is required. The present paper will develop the automatic differentiation based concurrent subspace optimization(AD-CSSO) in MDO. An automatic differentiation tool in FORTRAN(ADIFOR) will be employed to evaluate sensitivities. The use of exact function derivatives in GSE, OSA and COP makes Possible to enhance the numerical accuracy during the iterative design process. The paper discusses how much influence on final optimal design compared with traditional all-in-one approach, finite difference based CSSO and AD-CSSO applying coupled design variables.

• The Journal of the Acoustical Society of Korea
• /
• v.20 no.1E
• /
• pp.7-13
• /
• 2001

In this paper we propose a new subspace tracking algorithm based on the PASTd (Projection Approximation Subspace Tracking with deflation). The algorithm is obtained via introducing the variable forgetting factor which adapts itself to the time-varying subspace environments. The tracking capability of the proposed algorithm is demonstrated by computer simulations in an abruptly changing DOA scenario. The estimation results of the variable forgetting factor PASTd(VFF-PASTd) outperform those of the PASTd in the nonstationary case as well as in the stationary case.

• Journal of Information Technology Applications and Management
• /
• v.23 no.1
• /
• pp.45-59
• /
• 2016

An ensemble classifier is a method that combines output of multiple classifiers. It has been widely accepted that ensemble classifiers can improve the prediction accuracy. Recently, ensemble techniques have been successfully applied to the bankruptcy prediction. Bagging and random subspace are the most popular ensemble techniques. Bagging and random subspace have proved to be very effective in improving the generalization ability respectively. However, there are few studies which have focused on the integration of bagging and random subspace. In this study, we proposed a new hybrid ensemble model to integrate bagging and random subspace method using genetic algorithm for improving the performance of the model. The proposed model is applied to the bankruptcy prediction for Korean companies and compared with other models in this study. The experimental results showed that the proposed model performs better than the other models such as the single classifier, the original ensemble model and the simple hybrid model.

• ETRI Journal
• /
• v.31 no.2
• /
• pp.193-200
• /
• 2009

In this paper, we study the performance of subspace-based multiuser detection techniques for multicarrier code-division multiple access (MC-CDMA) systems. We propose an improvement in the PASTd algorithm by cascading it with the classical Gram-Schmidt procedure to orthonormalize the eigenvectors after their sequential extraction. The tracking of signal subspace using this algorithm, which we call OPASTd, has a faster convergence as the eigenvectors are orthonormalized at each discrete time sample. This improved PASTd algorithm is then used to implement the subspace blind adaptive multiuser detection for MC-CDMA. We also show that, for multiuser detection, the complexity of the proposed scheme is lower than that of many other orthogonalization schemes found in the literature. Extensive simulation results are presented and discussed to demonstrate the performance of the proposed scheme.

• Journal of the Korean Institute of Telematics and Electronics C
• /
• v.35C no.6
• /
• pp.88-98
• /
• 1998

Subspace classifier is a popular pattern recognition method based on Karhunen-Loeve transform. This classifier describes a high dimensional pattern by using a reduced dimensional subspace. Because of the loss of information induced by dimensionality reduction, however, a subspace classifier sometimes shows unsatisfactory recognition performance to the patterns having quite similar principal components each other. In this paper, we propose the use of multiple novelty neural network classifiers constructed on novelty vectors to adopt minor components usually ignored and present a method of improving recognition performance through combining those with the subspace classifier. We develop the proposed classifier on handwritten numeral database and analyze its properties. Our proposed classifier shows better recognition performance compared with other classifiers, though it requires more weight links.

• Transactions of the Korean Society of Mechanical Engineers A
• /
• v.27 no.4
• /
• pp.551-558
• /
• 2003

An efficient and reliable subspace iteration algorithm using the block algorithm is proposed. The block algorithm is the method dividing eigenpairs into several blocks when a lot of eigenpairs are required. One of the key for the faster convergence is carefully selected initial vectors. As the initial vectors, the proposed method uses the modified Ritz vectors for guaranteering all the required eigenpairs and the quasi-static Ritz vectors for accelerating convergency of high frequency eigenvectors. Applying the quasi-static Ritz vectors, a shift is always required, and the proper shift based on the geometric average is proposed. To maximize efficiency, this paper estimates the proper number of blocks based on the theoretical amount of calculation in the subspace iteration. And it also considers the problems generated in the process of combining various algorithms and the solutions to the problems. Several numerical experiments show that the proposed subspace iteration algorithm is very efficient, reliable ,and accurate.