• Proceedings of the KIEE Conference
• /
• 1997.11a
• /
• pp.594-596
• /
• 1997

This paper presents the speaker-dependent Korean isolated digit recognition system using advanced competitive learning. Since competitive learning algorithms are easy and simple to implement, they are used in various fields. The proposed recognition algorithm consists of three procedures: comparing winning number of codebook vectors, selecting the representative vector out of codebook vectors, and generating a new codebook with the representative vectors. In this paper, we use a sound blaster 16 for obtaining speech data. Speech data are sampled by 16 bits and 11 kHz sampling rate.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2009.01a
• /
• pp.238-241
• /
• 2009

In order to design a good quantizer for the underlying distribution using a training sequence (TS), the traditional approach is seeking for the empirical minimum based on the empirical risk minimization principle. As the size of TS increases, we may obtain a good quantizer for the true distribution. However, if we have a relatively small TS, searching the empirical minimum for the TS causes the overfitting problem, which even worsens the performance of the trained codebook. In this paper, the performance of codebooks trained by small TSs is studied, and it is shown that a piecewise uniform codebook can be better than an empirically minimized codebook is.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.8 no.4
• /
• pp.1292-1306
• /
• 2014

This paper proposes a space-polarization division multiple access (SPDMA) system that has limited feedback channels. The system simultaneously serves data streams to multiple mobile users through dual-polarized antenna arrays, by using pre-determined sets of precoding vectors that are orthogonal in both space and polarization domains. To this end, a codebook whose elements are sets of the precoding vectors is systematically designed based on the discrete Fourier transform (DFT) matrix and considering the power imbalance of polarized channels. Throughput of the SPDMA system is evaluated and compared to that of space division multiple access (SDMA) system, according to the various parameters including cross polarization discrimination (XPD). The results show that the throughput of SPDMA system outperforms that of SDMA in the environments of high XPD with many mobile users.

• The Transactions of the Korea Information Processing Society
• /
• v.7 no.11
• /
• pp.3462-3469
• /
• 2000

In this paper,new image vector quantization method for improvemtnt computation cost and compression ratio is proposed. A proposed method could saved the cornputatio cost of codebook eneration and encoding using partial codebook search, partial codevector elements, and interuplion criterion. And to improve cornpression ratio of codegook index lossless coding, codebook rearrangement, and variable length coding scheme are used.

• Journal of the Korean Institute of Telematics and Electronics B
• /
• v.33B no.7
• /
• pp.107-115
• /
• 1996

The average distortio of the vector quantizer is calcualted using a probability function F of the input source for a given codebook. But, since the input source is unknown in geneal, using the sample vectors that is realized from a random vector having probability function F, a time-average opeation is employed so as to obtain an approximation of the average distortion. In this case the size of the smple set should be large so that the sample vectors represent true F reliably. The theoretical inspection about the approximation, however, is not perfomed rigorously. Thus one might use the time-average distortion without any verification of the approximation. In this paper, the convergence characteristics of the time-average distortions are theoretically investigated when the size of sample vectors or the size of codebook gets large. It has been revealed that if codebook size is large enough, then small sample set is enough to obtain the average distortion by approximatio of the calculated tiem-averaged distortion. Experimental results on synthetic data, which are supporting the analysis, are also provided and discussed.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2001.10a
• /
• pp.579-582
• /
• 2001

Vector Transformation is a new method unified vector quantization and coding. So far, codebook generation applied to coding was LBG algorithm. But using the advantage of SOFM(Self-Organizing Feature Map) based on neural network can improve a system's performance. In this paper, we generated VTC(Vector Transformation Coding) codebook applied with SOFM algorithm and compare the result for several coding rates with LBG algorithm. The problem of Vector quantization is complicated calculation and codebook generation. So, to solve this problem, we used neural network approach method.

• Journal of the Institute of Electronics Engineers of Korea TC
• /
• v.45 no.12
• /
• pp.20-25
• /
• 2008

Codebook-based opportunistic beamforming (COBF) technique provides a beam selection diversity to the conventional opportunistic beamforming. In this paper, we design the random matrix and codebook for the COBF technique based on a kappa-factor channel model. Applying the proposed design to the COBF, nearly optimal beams are generated. Therefore, the COBF shows an outstanding performance without regard to the channel correlation related to the kappa-factor.

• Journal of Communications and Networks
• /
• v.9 no.2
• /
• pp.141-149
• /
• 2007

Multiple-input multiple-output (MIMO) transmit pre-coding/beamforming can significantly improve system spectral efficiency. However, several obstacles prevent precoding from wide deployment in early wireless networks: The significant feedback overhead, performance degradation due to feedback delay, and the large storage requirement at the mobile devices. In this paper, we propose a precoding method that addresses these issues. In this approach, only 3 or 6 bits feedback is needed to select a precoding matrix from a codebook. There are fifteen codebooks, each corresponding to a unique combination of antenna configuration (up to 4 antennas) and codebook size. Small codebooks are prestored and large codebooks are efficiently computed from the prestored codebook, modified Hochwald method and Householder reflection. Finally, the feedback delay is compensated by channel prediction. The scheme is validated by simulations and we have observed significant gains comparing to space-time coding and antenna selection. This solution was adopted as a part of the IEEE 802.16e specification in 2005.

• The Transactions of the Korea Information Processing Society
• /
• v.4 no.1
• /
• pp.149-160
• /
• 1997

This paper is concerned with the problem of speaker-adaptive speech synthes is method using a mapped codebook designed by fuzzy mapping on FLVQ (Fuzzy Learning Vector Quantization). The FLVQ is used to design both input and reference speaker's codebook. This algorithm is incorporated fuzzy membership function into the LVQ(learning vector quantization) networks. Unlike the LVQ algorithm, this algorithm minimizes the network output errors which are the differences of clas s membership target and actual membership values, and results to minimize the distances between training patterns and competing neurons. Speaker Adaptation in speech synthesis is performed as follow;input speaker's codebook is mapped a reference speaker's codebook in fuzzy concepts. The Fuzzy VQ mapping replaces a codevector preserving its fuzzy membership function. The codevector correspondence histogram is obtained by accumulating the vector correspondence along the DTW optimal path. We use the Fuzzy VQ mapping to design a mapped codebook. The mapped codebook is defined as a linear combination of reference speaker's vectors using each fuzzy histogram as a weighting function with membership values. In adaptive-speech synthesis stage, input speech is fuzzy vector-quantized by the mapped codcbook, and then FCM arithmetic is used to synthesize speech adapted to input speaker. The speaker adaption experiments are carried out using speech of males in their thirties as input speaker's speech, and a female in her twenties as reference speaker's speech. Speeches used in experiments are sentences /anyoung hasim nika/ and /good morning/. As a results of experiments, we obtained a synthesized speech adapted to input speaker.

• Journal of the Institute of Electronics and Information Engineers
• /
• v.52 no.7
• /
• pp.22-30
• /
• 2015

For long-term evolution (LTE) MIMO transmission, codebooks are used to utilize the estimated channel information under the limited feedeback environment, and related study has been actively performed. Existing codebooks include codevectos constructed based on vector quantization (VQ) and discrete Fourier transform (DFT), and the LTE standard specifies codebooks modified from these examples to support up to 8 transmit antennas. As the number of antennas increases and as the spatial channel model is used as a standard environment to evaluate the LTE transmission performance, new beamforming methods as well as codebook designs are needed. In this paper, we implement the 3-dimensional spatial channel model (3D-SCM) to analyze the key statistical characteristics of the generated channel, and present efficient ways of determining corresponding codebooks. In particular, we propose a nonuniform-phase DFT-based codebook to improve the existing uniform-phase DFT-based codebook, and evaluate its performance under the given SCM transmission environment. There exists a strong tendancy in statistical distributions of the phase difference between adjacent antenna elements for the SCM, which can be appropriately exploited in codebook design to produce a performance gain over the existing design.