Adaptive decision feedback equalizer for SISO communication channel using combined FIR-neural network and fast block LMS algorithm

Abstract

In this paper, an adaptive nonlinear decision feedback equalizer(DFE) is designed using three blocks that is feed forward filter, feedback filter and single layer neural network architecture with functional expansions of input patterns. Fast Block Least Mean Square (FBLMS) is used to update coefficients of all filters used in the design and the algorithm is based on block processing using FFT and IFFT which substantially reduces Bit Error. Thus the adaptive DFE as a part of the demodulator will improve the error performance of the receiver. The performance of the DFE is evaluated in terms of BER and eye diagrams for the BPSK modulated transmitted data sequence. Non-linear minimum phase time invariant channel model is used for Single Input Single Output (SISO) communication systems which introduces ISI in the modulated data stream. The comparison results are obtained by using other coefficients update algorithms like LMS and Block LMS in presence of additive white Gaussian noise.