Fractionally spaced coherent detection with DF MAP estimation of statistically known time-varying Rayleigh fading channel

Coherent symbol by symbol detection using fractionally spaced samples in a time-varying frequency flat Rayleigh fading channel is presented. The proposed detector incorporates a decision feedback (DF) maximum a posteriori (MAP) channel estimator that can be applied to a time-varying flat Rayleigh fading channel with an arbitrary but known spaced-time correlation function. The best achievable bit error rate performance for this detector is obtained by assuming correct bit decisions while performing channel estimation. Differential encoding is used in order to prevent the phase reversal phenomenon during the estimator's operation in a decision directed mode. Numerical results suggest that with only a few samples per symbol and with a short memory estimator depth, most of the gain due to oversampling can be achieved.