Application of a Transform-Domain Adaptive Algorithm to the Design of Equalizers in Digital Communication Systems

• Main Authors: Wu, Shuenn-Tay, 吳順泰
• Other Authors: Ching-Hsiang Tseng
• Format: Others
• Language: zh-TW
• Published: 1997
• Online Access: http://ndltd.ncl.edu.tw/handle/22989290138781938824

碩士 === 國立海洋大學 === 電機工程學系 === 85 === An equalizer is a structure used in digital communication systems foreliminating the intersymbol interference(ISI) generated in the channel. Itsperformance is usually nastered by the adopted adaptive algorithm. Therefore, equalizers using different algorithms may have different limitations on their applications. Every conventional adaptive algorithm may have its own disadvantage,such as low convergence rate(e.g., LMS), high computational complexity, and numerical instability(e.g., RLS). Aiming ata these shortcomings, we propose atransform- domain approximate RLS(TARLS) algorithm. The proposed methodis shown to be behave similarly to the RLS in terms of convergence rate whilekeeping its computational complexity to be of the same order as that of LMS.Besides, we also consider the performance of the proposed method under the environment of finite precision. We find that the proposed method can be immune from the possible numerical instability introduced by the quantization error. The proposed method is also used to replace LMS type algorithm employed in a blind equalizer. Such a combination is proven to be able to improvenot only the low convergence rate of the blind equalizer but also the numericalinstability caused by finite precision. In addition, in view of the fact that a simple structure is often preferred for real-time implementation of equalizers,we implement some LMS type and blind equalizers by using a digital signal processor( DSP). Real-time processing of data signals is achieved in all cases. We use computer simulation to verify that the proposed method, whenbeing applied to a system identification task can indeed perform better than theLMS and RLS algorithm do. By simulating the QPSK signal passing through a multipath fading channel, the superiority of the por posed method over the conventional method is demonstrated. At last, by using a signal processingsoftware, we demonstrate the feasibility of using digital signal processors toimplement real-time processing adaptive equalizers.