Co-channel and Adjacent-channel Interference Cancellation Using Neural Networks

by

Xianrong Yao

M. Eng., 1993

Abstract

This thesis investigates the problem of adaptive equalization in the presence of intersymbol interference, additive white Gaussian noise, co-channel and adjacent channel interference. The multilayer precetron-based decision feedback equalizer (MLPDFE) is designed to realize a sophistical non-linear adaptive equalizer capable of operating under poor signal to interference ratio, poor signal to noise ratio and server timing misalignment of co-channels with respect to the main channel conditions.

This thesis starts by a demonstration of the sensitivity of the MLPDFE performance (in terms of averaged MSE) on its perception configuration and adaptive parameters. The effort of co-channel and adjacent-channel interference on the system performance is investigated separately. The results of a computer simulation study to compare the performance of the MLPDFE with that of the least mean square decision feedback equalizer (LMSDFE) in terms of the SER with a QPSK signaling are presented. The results indicate that the MLPDFE provides better SER performance relative to the LMSDFE when only co-channel interference presents, especially in server co-channel timing misalignment conditions. Also, in the presence of only adjacent-channel interference, the MLPDFE has very similar SER performance as that of the LMSDFE.

Back to the list of graduate students