On binary DPSK detection

A discussion of optimum-detector structure indicates that optimum detection of binary DPSK requires a frequency reference but gives a bit error probability (Pe) in the absence of ISI, expressed by (1/2)exp(-Eb/eta), where Eb is the bit energy and eta/2 is the amplitude of the power spectrum of the additive Gaussian noise. If a narrowband IF filter is used, the bandwidth should be chosen as a carrier frequency equaling 0.57/T (T = bit time) for the best performance, in which case ISI is nearly eliminated, and 1 dB degradation with respect to the optimum detector in Pe performance results. For other IF bandwidths, Pe becomes worse or more ISI results, or both. For a wideband IF and integrate-and-dump-detector there is essentially no ISI but there is 2-dB degradation with respect to the optimum detector in performance at a high SNR.