On Predicting Convergence of Iterative MIMO Detection-Decoding With Concatenated Codes

We evaluate the applicability of methods from stochastic decoding analysis to convergence prediction of iterative multiple-input-multiple-output (MIMO) detection decoding. The one-parametric conditional Gaussian log-likelihood ratio (LLR) distribution model, which underlies EXtrinsic Information Transfer (EXIT) charts, is not adequate for some practically relevant scenarios such as fading MIMO channels. A more recent two-parametric Gaussian model, which better fits arbitrary distributions, can be combined with an offset compensation to allow for a chart-based prediction of the convergence of iterative receiver processing in these cases.