Least Bit Error Rate Adaptive Nonlinear Equalizers for Binary Signalling

Download

	Postscript Download (559Kb)
	PDF Download (535Kb)

Description/Abstract

The paper considers the problem of constructing adaptive minimum bit error rate (MBER) neural network equalizers for binary signalling. Motivated from a kernel density estimation of the bit error rate (BER) as a smooth function of training data, a stochastic gradient algorithm called the least bit error rate (LBER) is developed for adaptive nonlinear equalizers. This LBER algorithm is applied to adaptive training of a radial basis function (RBF) equalizer in a channel intersymbol interference (ISI) plus co-channel interference setting. Simulation study shows that the proposed algorithm has a good convergence speed and a small-size RBF equalizer trained by the LBER can closely approximate the performance of the optimal Bayesian equalizer. The results also demonstrates that the standard adaptive algorithm, the least mean square (LMS), performs poorly for neural network equalizers, due to the reason that the minimum mean square error (MMSE) is irrelevant to the equalization goal.