Comparison of Advanced Detection Techniques for QPSK Signals in Sub-Nyquist WDM Systems

We compare the performance of detecting quadrature-phase-shift-keying (QPSK) in a sub-Nyquist channel spacing system as a QPSK and quaternary duobinary (QDB) signal using maximum likelihood sequence estimation (MLSE) and maximum a posteriori (MAP) criterion with both simulations and experiments. Three different detection schemes are studied: 1) QPSK detection with a multi-tap linear equalizer followed by a MLSE or MAP detector; 2) QPSK detection with only a MLSE or MAP detector; 3) QDB detection with a MLSE or MAP detector, where a multi-tap linear equalizer and a delay-&-add operation are placed before the MLSE/MAP detector to convert a QPSK signal to a QDB signal. In simulations, the performance of a 112-Gb/s polarization-division-multiplexed (PDM) QPSK signal using the three detection schemes is investigated with different MLSE/MAP tap numbers and WDM channel spacings. In experiments, the performance of a three channel 112-Gb/s PDM-QPSK at a 25-GHz channel spacing is examined. Both simulations and experiments show that the QDB detection with multi-tap MLSE does not provide better performance than that with 2-tap MLSE, and the QDB detection with 2-tap MLSE has better filtering tolerance than the QPSK detection with linear equalization followed by a multi-tap MLSE/MAP detector.