Turbo DFE Assisted Time-Frequency Packing for Probabilistically Shaped Terabit Superchannels

Abstract

We present a probabilistically shaped (PS) time-frequency-packing (TFP) wavelength-division multiplexing superchannel system employing higher order modulation (HoM) formats, with an objective to improve the spectral efficiency (SE). However, TFP systems suffer from inter-symbol interference (ISI) and/or inter-carrier interference (ICI). Additionally, the bandwidth limitations of the wavelength selective switches in the fiber link may cause severe ISI for the edge sub-channels (SCs) in a superchannel. Mitigation of such interferences is particularly challenging for HoM systems, since the implementation of the well-known turbo equalization schemes, such as the Bahl-Cocke-Jelinek-Raviv equalizer, is computationally challenging for larger constellations. In this paper, we investigate an expectation propagation based, computationally efficient, turbo decision feedback equalizer for ISI cancellation, in tandem with a parallel interference cancellation based ICI mitigation. By optimizing the parameters in the shaping and the packing dimensions, we show through our numerical results that the proposed PS-TFP superchannels enabling 1.8 Tbps data rates offer up to 1.05 dB packing gain over an unpacked system using the same modulation format, and a 1.15 dB shaping gain over an unshaped system that uses a lower modulation order to achieve the same SE.

Keywords

• Faster-than-Nyquist (FTN)
• time-frequency packing (TFP)
• probabilistic shaping (PS)
• decision feedback equalization (DFE)
• expectation propagation (EP)
• superchannel