Modified EMLMAP Algorithm Using Variable Scaling Factor for Low-Power Implementation of Turbo Decoder

Abstract

We propose a modified enhanced max-log-MAP (EMLMAP) algorithm using variable scaling factors (SFs) for low-power implementations of a Turbo decoder. We find the sub-optimum SFs ^® which are quantized from 0.6 to 0.9 using extrinsic transfer characteristics (EXIT) analysis. Based on the analysis, the variation scheme of the SF is described, which is a function of half-iteration hard-decision-aided (HIHDA) early termination criteria S. Using computer simulations applying 8-state WCDMA Turbo codes of the rate R=1/3 on memoryless binary-input continuous-output AWGN channel, we analyze the BER performance and average number of iterations of the proposed algorithm by comparing that of the conventional EMLMAP algorithm which is applied fixed SF and HI-HDA early termination scheme. According to results of the simulations, the proposed algorithm has low-power effciency and improvement in BER performance for a large interleaver of length N>3000.