Eunchul Yoon

MIMO capacity with channel uncertainty: does feedback help?

We investigate ergodic capacities and optimal transmitter strategies in Rayleigh fading multiple input multiple output (MIMO) channels with spatial correlation, when there exist channel uncertainties arising from the combined effect of channel estimation error and limited feedback. We consider both covariance feedback and instantaneous feedback, and formulate optimization problems that determine the capacities and optimal transmitter designs for both cases. In the high SNR regime, the optimal solutions have simple closed form formulas that involve inverting the channel covariance and waterfilling over instantaneous channel gains. Numerical results show that instantaneous feedback gives large capacity gain at low SNR and is also helpful at high SNR. Covariance feedback, on the other hand, seems to give little gain at mid SNR, but is almost as good as instantaneous feedback at high SNR under a reasonable channel estimation quality.

An efficient mining algorithm for maximal weighted frequent patterns in transactional databases

In the field of data mining, there have been many studies on mining frequent patterns due to its broad applications in mining association rules, correlations, sequential patterns, constraint-based frequent patterns, graph patterns, emerging patterns, and many other data mining tasks. We present a new algorithm for mining maximal weighted frequent patterns from a transactional database. Our mining paradigm prunes unimportant patterns and reduces the size of the search space. However, maintaining the anti-monotone property without loss of information should be considered, and thus our algorithm prunes weighted infrequent patterns and uses a prefix-tree with weight-descending order. In comparison, a previous algorithm, MAFIA, exponentially scales to the longest pattern length. Our algorithm outperformed MAFIA in a thorough experimental analysis on real data. In addition, our algorithm is more efficient and scalable.

Space-Frequency Precoding with Space-Tap Correlation Information at the Transmitter

In closed-loop methods for obtaining exact channel state information at the transmitter (CSI-Tx), the overhead associated with the feedback can be excessive for fast mobiles. Channel statistics-based CSI-Tx requires a much smaller overhead and is, therefore, attractive for use with fast mobiles. We study ways to exploit correlation-based CSI-Tx in a multiple-input multiple-output (MIMO)-orthogonal frequency-division multiplexing (OFDM) system. We focus on a channel environment in which spatial and tap correlations are present. We propose a channel model for the case that spatial and tap correlations can be separated and show that in this case channel correlation decreases the ergodic capacity of an MIMO-OFDM system when no CSI-Tx is available. However, this decrease can be mitigated when correlation-based CSI-Tx is exploited. We introduce an optimal precoding approach to maximize capacity with spatial and tap correlation-based CSI-Tx. We also propose a statistical waterfilling scheme, which leads to almost optimal capacity performance without requiring computationally intensive numerical optimization. Based on these approaches, the impact of spatial and tap correlations is investigated.

Space-frequency precoding for an OFDM based system exploiting spatial and path correlation

We investigate the capacity behavior of an OFDM wireless link for the frequency selective MIMO channel with correlated paths. The derivation is based on a particular channel model that characterizes path correlation as well as spatial correlation. As in the case of spatial correlation, path correlation can reduce capacity. If there is no channel knowledge at the transmitter, the capacity reduction of an OFDM based system due to path correlation can be explained in terms of the effective SNR at each tone. With spatial and path correlation information at the transmitter, we derive the optimal space-frequency precoder at every tone by numerical optimization techniques. We propose a sub-optimal precoding scheme with lower complexity which leads to a tight lower bound of the capacity of the OFDM system.

Efficient Mining of Robust Closed Weighted Sequential Patterns Without Information Loss

Sequential pattern mining has become one of the most important topics in data mining. It has broad applications such as analyzing customer purchase data, Web access patterns, network traffic data, DNA sequencing, and so on. Previous studies have concentrated on reducing redundant patterns among the sequential patterns, and on finding meaningful patterns from huge datasets. In sequential pattern mining, closed sequential pattern mining and weighted sequential pattern mining are the two main approaches to perform mining tasks. This is because closed sequential pattern mining finds representative sequential patterns which show exactly the same knowledge as the complete set of frequent sequential patterns, and weight-based sequential pattern mining discovers important sequential patterns by considering the importance of each sequential pattern. In this paper, we study the problem of mining robust closed weighted sequential patterns by integrating two paradigms from large sequence databases. We first show that the joining order between the weight constraints and the closure property in sequential pattern mining leads to different sets of results. From our analysis of joining orders, we suggest robust closed weighted sequential pattern mining without information loss, and present how to discover representative important sequential patterns without information loss. Through performance tests, we show that our approach gives high performance in terms of efficiency, effectiveness, memory usage, and scalability.

A blog ranking algorithm using analysis of both blog influence and characteristics of blog posts

In recent years, with the increasing number of blogs to share information, the ratio of blogs on the World Wide Web has been raised. In this regard, the problem of information quality has come up due to the rapidly increasing amount of information in a blogosphere. Therefore, discovering good quality information is one of the significant issues in the blog space. In this paper, we propose an algorithm for efficient blog ranking, called WCT (a blog ranking algorithm using Weighted Comments and Trackbacks). This method performs a ranking process through not only interconnection analysis of blogs but also structural weights for contents in the blogs. Moreover, we conduct performance evaluation and discuss the performance between our algorithm and a previous algorithm by comparing their experimental results, which show that our approach has higher performance than that of the other blog retrieval method.

Linear Precoding for High-K-Factor Channels Exploiting Channel Mean and Covariance Information

In this paper, we consider the high-K-factor Ricean channel, where the channel mean is of arbitrary rank, and the fading channel components are correlated. We propose two suboptimal linear precoding schemes based on the knowledge of the channel mean and covariance at the transmitter. The first precoder is derived from an approximate ergodic capacity expression for the high-K-factor Ricean channels, whereas the second is derived from Jensens inequality. We compare the capacity obtained to the true ergodic capacity found by using a numerical convex optimization algorithm. We further investigate the impact of the fading channel covariance, the rank of the channel mean, and the K-factor on the precoding schemes and the achievable ergodic capacity limit.

Subcarrier and power allocation for an OFDMA uplink based on tap correlation information

The impact of tap correlation on the achievable average sum rate of an OFDMA uplink is investigated. Tap correlation between channel taps was showed to reduce the average sum rate of an OFDMA uplink. However, if available subcarriers and power are allocated to multiusers based on their tap correlation information, a hefty portion of inherent multiuser diversity can be exploited in a cost effective manner leading to substantial improvement in spectral efficiency compared with uniform subcarrier and power allocation. The performance gain of such statistical adaptation is further investigated as a function of the number of simultaneously scheduled users per OFDM symbol.

A Time-Reversal-Based Transmission Using Predistortion for Intersymbol Interference Alignment

A time-reversal (TR)-based transmission using pre-distortion of the transmitted waveforms for intersymbol interference (ISI) alignment is proposed. This scheme differs from the previous TR-based pre-filtering schemes in that it uses distinctively designed pre-distorted waveforms that are based on the transmitted symbol information. In the proposed pre-distortion scheme, the successively received waveforms carrying adjacent transmitted symbols are aligned so that the received symbol power can be intensified. Since the overlapped portions of the successively transmitted waveforms from the proposed pre-distortion scheme are correlated, the power of the overall transmitted signal that is formed by time-shifting and adding the pre-distorted waveforms becomes larger than the sum of the individual pre-distorted waveform powers. In order to adjust the pre-distorted waveform power without measuring the actual power of the overall transmitted signal, an upper bound on the average power of the overall transmitted signal was derived analytically under a low SNR assumption. Simulation showed that the proposed pre-distortion scheme outperformed both conventional TR and the minimum mean square error (MMSE)-based pre-filtering when the ISI was serious with a high transmission rate.

Damped window based high average utility pattern mining over data streams

Abstract Data mining methods have been required in both commercial and non-commercial areas. In such circumstances, pattern mining techniques can be used to find meaningful pattern information. Utility pattern mining (UPM) is more suitable for evaluating the usefulness of patterns. The method introduced in this paper employs the high average utility pattern mining (HAUPM) approach, which is one of the UPM approaches and discovers interesting patterns of which the items have more meaningful relations among one another by using a novel utility measure. Meanwhile, past research on pattern mining algorithms mainly focus on mining tasks processing static database such as batch operations. Most continuous, unbounded stream data such as data constantly produced from heart beat sensors should be treated differently with respect to importance because up-to-date data may have higher influence than old data. Therefore, our approach also adopts the concept of the damped window model to gain more useful patterns in stream environments. Various experiments are performed on real datasets in order to demonstrate that the designed method not only provides important, recent pattern information but also requires less computational resources such as execution time, memory usage, scalability and significant test.