Filtered Carrier Phase Estimator for High-Order QAM Optical Systems

We investigate, using Monte Carlo simulations, the performance characteristics and limits of a low-complexity filtered carrier phase estimator (F-CPE) in terms of cycle-slip occurrences and signal to noise ratio (SNR) penalties. In this work, the F-CPE algorithm has been extended to include modulation formats whose outer-ring symbols have a quadrature phase shift keying (QPSK) symmetry, and which are applicable to metro and long-haul optical networks: QPSK, 8, 16, and 64 quadrature-amplitude modulation (QAM). The proposed joint-polarization approach, where the number of non-null symbols in a simplified QPSK partition is increased, shows a further improvement in robustness against cycle slips, resulting in cycle-slip-free operation at symbol rate 32 GBd and laser linewidths up to 900 kHz, for the range of investigated SNRs. In addition, it reduces SNR penalties for only a small incremental complexity. We also propose a method for constellation alignment that exploits F-CPE computational blocks to minimize the electronic footprint, in order to compensate for an arbitrary rotation, introduced by F-CPE. We further show that F-CPE can successfully recover the phase of a 32QAM signal that does not have the outer-ring QPSK symmetry, albeit with increased penalties and higher cycle-slip rates. A performance comparison between F-CPE, the blind phase search algorithm, and a data-aided phase estimator, is also presented.