Improved EXIT Analysis for Turbo Decoding

摘要

Extrinsic information transfer (EXIT) chart is a useful tool for analyzing iterative decoding behavior. One crucial challenge in realizing it is that a precise estimate of the probability density function (PDF) of the extrinsic information at the output of a soft-input soft-output decoder must be made to obtain mutual information. To achieve this, we employed a kernel density estimator (KDE), as opposed to using the traditional histogram estimator, when a fairly small interleaver size is available. Simulation results showed that the KDE approach outperformed the histogram method for estimating the PDF in the EXIT band analysis in terms of bit error rate prediction.