Semi-blind DFE for MIMO Optical Communications

In the last decade, the coherent transmission over fiber has attracted many research interests, mainly for long-haul transmission, due to its advantages in terms of DSP processing. However, with the raise of demand for higher baud-rates and increased technical requirements, the current DSP algorithms have failed to offer suitable performance in the extreme transmission scenarios such as high State-of-Polarization (SOP) frequencies. Meanwhile, the joint effect of polarization mode dispersion (PMD), fast SOP rotation, carrier frequency offset (CFO) and phase noise significantly degrade the performance of the whole system. In this scenario, the use of conventional blind equalizers such as constant modulus algorithms (CMA) and multi-modulus algorithms (MMA) would not be enough for symbol detection. In this work, we propose the decision feedback estimation (DFE) as the baseline for a joint Multiple-Input Multiple-Output (MIMO) equalization and CFO compensation. The devised DFE-like algorithm, implemented in an adaptive manner, allows MIMO equalization even at high CFO and SOP rates. The algorithm is developed in both blind and semi-blind scenarios based on QAM using the probabilistic constellation shaping (PCS) technique.