Improving differential detection of MDPSK by noise prediction

A new technique is proposed to improve the performance of differential detection of MDPSK (M-ary differential phase-shift keying) significantly by reducing the noise in the decision variable by means of noise prediction. A nonlinear time-variant FIR or IIR structure is used for the prediction-error filter. For both filter structures the optimum filter coefficients are derived, assuming transmission over an additive white Gaussian noise (AWGN) channel. For the IIR structure a setting for the filter coefficient exists which yields a high performance for both AWGN and flat fading (Rayleigh and Rician) channels. Moreover, a simple decision rule is derived for the IIR structure which allows an implementation with very low computational complexity