Design Rules for Adaptive Turbo Equalization in Fast-Fading Channels

This paper uses the mutual information (MI) transfer metric to characterize the behavior of an adaptive turbo equalization scheme. We consider recursive systematic convolutional coding at the input to a multipath fastfading channel. The equalizer is a recent adaptive, nearoptimal APP algorithm that incorporates MMSE channel estimation. We develop design rules for this turbo equalizer using EXIT charts. We optimize the use of pilot bits and determine the best ratio of code bits to pilot bits, for fixed data and symbol rates.