Study on the Convergence of the Generalized Decision Feedback Equalizer Algorithm with Error Feedback Filter

In digital communications, a time-dispersive variation channel causes inter-symbol interference (ISI) and distorts the received signal, thus degrading the system performance. The distortion caused by ISI may be mitigated by passing the received signal through an adaptive equalizer such as linear equalizer, maximum likelihood sequence estimator (MLSE) and decision feedback equalizer (DFE). This paper demonstrates a new approach to improving the convergence of the DFE algorithm with error feedback filter in order to reduce the correlation of error signals and also to decrease the residual error variance. The paper theoretically analyzes the convergence of generalized DFE with error feedback filter, and proves that the minimum mean square error (MMSE) monotonically decreases and stably converges when the order of the error feedback filter increases. The simulation clearly shows that the proposed DFE algorithm results in better ERLE (Echo Return Loss Enhancement) than previous ones and particularly the significant improvement of BER performance in case that the number of taps in the error feedback filter increases under low SNR environment.