Blind Adaptive Channel Equalization with Performance Analysis
- PDF / 452,972 Bytes
- 9 Pages / 600.03 x 792 pts Page_size
- 47 Downloads / 213 Views
Blind Adaptive Channel Equalization with Performance Analysis Shiann-Jeng Yu1 and Fang-Biau Ueng2 1 National
Center for High Performance Computing, No. 21 Nan-Ke 3rd Road, Hsin-Shi, Tainan County 744, Taiwan of Electrical Engineering, National Chung-Hsing University, 250 Kuo-Kuang Road, Taichung 402, Taiwan
2 Department
Received 4 March 2005; Revised 25 August 2005; Accepted 26 September 2005 Recommended for Publication by Christoph Mecklenbr¨auker A new adaptive multiple-shift correlation (MSC)-based blind channel equalizer (BCE) for multiple FIR channels is proposed. The performance of the MSC-based BCE under channel order mismatches due to small head and tail channel coefficient is investigated. The performance degradation is a function of the optimal output SINR, the optimal output power, and the control vector. This paper also proposes a simple but effective iterative method to improve the performance. Simulation examples are demonstrated to show the effectiveness of the proposed method and the analyses. Copyright © 2006 Hindawi Publishing Corporation. All rights reserved.
1.
INTRODUCTION
Traditional adaptive equalizers are based on the periodic transmission of a known training data sequence in order to identify or equalize a distorted channel with intersymbol interference (ISI). However, the use of training data sequence may be very costly in some applications. Blind channel equalizers (BCE) without training data available receive much attention in recent years [1–15]. Early blind equalization techniques [1, 2] exploited the higher order statistics (HOS) of the output to identify the channels. Unfortunately, the HOS-based BCE requires a large number of data samples and huge computation load which limit their applications in fast changing environments. To circumvent the shortcomings of the HOS-based approaches, second-order statistics (SOS) was considered in BCE. The SOS-based BCE was developed based on cyclostationary characteristics of the signal. The first SOS-based BCE was derived by Tong et al. [3]. They demonstrated that the SOS is sufficient for blind adaptive equalization by using fractionally sampling or using an array of sensors. Since that, extensive researches were explored in the literature. The well-known approaches are the least-squares, the subspace, and the maximum likelihood [3, 8, 9]. These blind equalizers were termed the two-step methods which estimate multiple channel parameters first and then equalize the channels based on the estimated channel parameters. However, the two-step methods are not optimal because they do not take the channel estimation error into account in the
second-step optimization procedure. Recently, direct equalization estimators become more attractive [10–13]. The linear prediction-based equalizer was developed by [13]. Work [12] used the adaptive beamforming technique to develop a constrained optimization method. Multiple-shift correlation (MSC) of the signals can be used in a partially adaptive channel equalizer to achieve fast convergence speed and low computation
Data Loading...
