A perceptron-based adaptive equalizer for PAM signals

Author

Giridhar, K. ; Shynk, John J.

Author_Institution

Dept. of Electr. & Comput. Eng., California Univ., Santa Barbara, CA, USA

Volume

3

fYear

1992

fDate

10-13 May 1992

Firstpage

1581

Abstract

An adaptive equalizer based on a multilayer perceptron and the backpropagation training algorithm is presented for M-ary PAM (pulse amplitude modulation) digital communication systems. The configuration is an extension of the conventional adaptive FIR (finite impulse response) equalizer whereby the linear filter is replaced by a nonlinear feedforward network with nodes that saturate at ±1. The training signal is derived by encoding the PAM signals via a bit-mapping strategy such that the output layer of the perceptron requires log₂ (M) nodes. A corresponding number of error signals is generated, and these are used to drive the backpropagation algorithm. Computer simulations are presented to illustrate the performance of the equalizer for two different bit mappings

Keywords

backpropagation; digital signals; equalisers; feedforward neural nets; pulse amplitude modulation; M-ary PAM; PAM signals; backpropagation training algorithm; bit-mapping strategy; digital communication systems; error signals; multilayer perceptron; nonlinear feedforward network; perceptron-based adaptive equalizer; training signal; Adaptive equalizers; Adaptive filters; Amplitude modulation; Backpropagation algorithms; Digital communication; Encoding; Finite impulse response filter; Multilayer perceptrons; Nonlinear filters; Pulse modulation;

fLanguage

English

Publisher

ieee

Conference_Titel

Circuits and Systems, 1992. ISCAS '92. Proceedings., 1992 IEEE International Symposium on

Conference_Location

San Diego, CA

Print_ISBN

0-7803-0593-0

Type

conf

DOI

10.1109/ISCAS.1992.230195

Filename

230195