Robust decision feedback detection of M-DPSK

Decision feedback differential detection (DFDD) of M-DPSK in Rayleigh fading channels significantly outperforms the conventional two-symbol differential detector. However, computation of the feedback coefficients of DFDD requires the knowledge of second order statistics of the channel, and any imperfections therein can severely degrade the error performance of DFDD. While adaptive DFDD schemes circumvent this problem, high computational complexity associated with them may prohibit their practical application. In this paper, we propose a "fixed-coefficient" (i.e., no adaptation is required) DFDD that is robust to variations in channel statistics, provided that the power spectral density of the channel varies-over a known set of functions. For the special case of Jakes\´ fading model, the proposed DFDD performs close to the recursive least square algorithm based DFDD if the Doppler spread is unknown to the receiver, providing excellent performance versus complexity trade-off.