Reduced-Complexity Equalizer Design

• Main Authors: Yu-Sheng Chen, 陳育聖
• Other Authors: Chien-Hsing Wu
• Format: Others
• Language: zh-TW
• Published: 2000
• Online Access: http://ndltd.ncl.edu.tw/handle/21907212384809887119

碩士 === 國立雲林科技大學 === 電子工程與資訊工程技術研究所 === 88 === This thesis aims at the development of various equalization methods for efficient VLSI realization, including 1. Time-domain equalization for asymmetric digital subscriber line (ADSL), 2. Reduced-complexity equalization via convolution decomposition (CD), and block processing, 3. Blind equalization based on LMS (least mean square error) and constant modulus (CM) criteria. We work out a new blind equalization method for acceleration of convergence based on the signed Godard algorithm (SGA) and decision feedback. The averaged stochastic gradient technique is also studied extensively to sustain a low-complexity design with improved convergence performance of the LMS method. For equalization in ADSL, we employ bisection search to select discrete step sizes adaptively for low-complexity VLSI implementation.