Improved Performance of the MMSE Linear Equalizer in Turbo Equalization by the use of Extrinsic Feedback Information

This paper presents a general structure for Minimum Mean Square Error (MMSE) equalizers from which we can derive instances such as the: two proposed Linear Equalizers with Extrinsic Feedback (LE-EF), Linear Equalizer (LE) and Decision Feedback Equalizer (DFE). The Extrinsic Information Transfer (EXIT) chart was used to analyze the two proposed MMSE LE-EF receivers and their characteristics was compared to the existing MMSE LE and Maximum A Posterior (MAP) equalizers used in turbo equalization. We show by using the Kolmogorov-Smirnov (K-S) test that the Gaussian assumption on the probability density function (PDF) of the a priori equalizer information after the first iteration is not valid. Simulation results demonstrate that the performance of one of the MMSE LE-EF receivers does approach the MAP equalizer and the MMSE linear equalizer with perfect extrinsic feedback (LE-PEF) for long block lengths at high signal to noise ratios (SNRs).