Data-driven efficient digital signal processing over a field trial space-division multiplexed fiber-optic transmission

This paper discusses the use of data-driven methodologies to improve multiple-input multiple-output (MIMO) Digital Signal Processing (DSP) performance for crosstalk cancelation in optical systems supporting multiple spatial modes. A crucial problem in time-domain equalization is the large quantity of data needed to compensate the effect of the crosstalk when the signals cover a large number of kilometers. In this respect, the study of techniques that are able to reduce the amount of information needed to perform equalization is addressed, and a reduction algorithm based on the Principal Components Analysis (PCA) and cross-correlation analysis to improve traditional equalizers' performance is proposed. Experimental validation is performed over the world-first space-division multiplexed (SDM) field trial.