خط مشی دسترسیدرباره ما
ثبت نامثبت نام
راهنماراهنما
فارسی
ورودورود
صفحه اصلیصفحه اصلی
جستجوی مدارک
تمام متن
منابع دیجیتالی
رکورد قبلیرکورد بعدی
Document Type : Latin Dissertation
Language of Document : English
Record Number : 151014
Doc. No : ET22806
Main Entry : SHAPAYAT ABRAR
Title Proper : NEURAL NETWORK BASED NONLINEAR BLIND EQUALIZATION SCECEMES FOR HIGH ORDER M-ary QAM SIGNALS
Note : This document is digital این مدرک بصورت الکترونیکی می باشد
Abstract : This t h k discusses some techniques to improve the neuml network based decision-di~ected blind epakation scheme. First, a Complex Back- Propagation algorithm isderived for Dud-Mode Modified Constant Modulus Algorithm; and to make it fine-tionat with MLP structureT a "gain-factor" is introduced to ovoid local minima.A new duaGmode scheme is also proposed, which switches Jkom blind to decision-directed mode mutlthout needing any gain-factor. Some new decision-di~ected b k dequalization algorithms are ako discussed. Second, an adaptive Stop-and-Go algo-ri۰hm is proposed to give better trucking capabzlify. Third, the idea of using twocascaded activation functions in the output neuron is proposed to obtain better cor-rzidzdzon between We reol and the imaginary parts of the output data and to get lowersteady-state Mean-Square-Emr. Fourth, the nonlinearity of the activation Jinc-tion is made adoptive based on the energy..,....-...,..tested for theQ1 PC1 bus cardBoth these projects mere sofixare des elopment efforts tonards contributing to dlfferentaspects of Roboucs and lZ1echatronics projects m the Controls and Roboucs Group..
Subject : Electericl tess
: برق
electronic file name : TL46032.pdf
Title and statement of responsibility and : NEURAL NETWORK BASED NONLINEAR BLIND EQUALIZATION SCECEMES FOR HIGH ORDER M-ary QAM SIGNALS [Thesis]
 
 
 
(در صورت عدم وضوح تصویر اینجا را کلیک نمایید)