A Low Computational-Complexity RBF-Assisted TEQ Based on Modified Bayesian Conditional PDF

In this paper, we propose a novel reduced-complexity Jacobian radial basis function (RBF)-assisted decision-feedback equalizer (DFE)-based turbo equalization (TEQ) scheme based on the modified Bayesian conditional probability density function. The proposed reduced-complexity Jacobian RBF DFE TEQ is capable of providing a near-identical performance to the Jacobian RBF DFE TEQ at a lower computational load in the context of both binary phase-shift keying (BPSK) and 4 quadrature amplitude modulation (QAM). When the channel impulse response (CIR) length is 3, in the subsequent iterations after the first iteration, the reduced-complexity Jacobian RBF DFE TEQ achieves the addition/subtraction reduction factors of 1.69 and 3.39 and the multiplication/division reduction factors of 1.14 and 2.46 when compared to the Jacobian RBF DFE TEQ for BPSK and 4 QAM, respectively. When the CIR length is 4, the addition/subtraction reduction factors of 3.6 and 14.9 and the multiplication/division reduction factors of 2 and 9.14 are attained for BPSK and 4 QAM, respectively. These are achieved by utilizing the a priori data provided from the channel decoder to reduce the number of the desired channel output states and to simplify each distance evaluation between each desired channel output state and the observed channel output vector in the subsequent iterations.