Likelihood-Based Selection Radius Directed Equalizer With Time-Multiplexed Pilot Symbols for Probabilistically Shaped QAM

Probabilistically shaped (PS) quadrature amplitude modulation (QAM) recently established itself as the solution to adopt for state-of-the-art coherent transponders. This in turn has drawn attention to the rework of conventional digital signal processing (DSP) algorithms, which are not optimal for QAM-PS. In this context, we propose a novel pilot-aided equalization technique based on the ubiquitous radius directed equalizer (RDE). Our solution employs both payload and time-multiplexed pilot symbols for the adaptive filters update. The payload symbols are treated on the basis of the likelihood of their correct blind assignment. We show that this approach is at the same time able to reduce the global pilot overhead (POH) required by the DSP chain and to provide greatly improved performance in the tracking ability of the equalizer in the event of fast state of polarization rotations. We describe a simulation environment in which we model accurately static and dynamic polarization-related impairments. We test our algorithm over different shaped modulation formats and in the presence of transceiver impairments and demonstrate its superior performance with respect to standard feed-forward implementations in all the scenarios considered.