Fig. 1. (a) Z-domain structure of the EFNS. (b) Linear model of clipping and quantization. (c) Z-domain structure of the EFNS based on a linear model.

Fig. 2. Quantization noise of the PSD.

Fig. 3. Constellations of the uniform signals and the PS signals.

Fig. 4. (a) Experimental setup of the IM/DD PAM4 transmission. (b) DSP flow in the transmitter. (c) DSP flow in the receiver. PRBS, pseudo-random binary sequence.

Fig. 5. (a) BER vs. PNOB when the ROP is −3 dBm. (b) BER vs. ROP when the PNOB is 3. (c) BER vs. ROP when the PNOB is 4.

Fig. 6. (a) Experimental setup of the coherent DP-16QAM transmission process. (b) DSP process at the transmitter. (c) DSP process at the receiver. CD, chromatic dispersion; CMA, constant modulus algorithm; CPR, carrier phase recovery; DD-LMS, decision directed least mean square; LO, local oscillator.

Fig. 7. (a) BER vs. PNOB when the OSNR is 30 dB. (b) BER vs. OSNR when the PNOB is 3. (c) BER vs. OSNR when the PNOB is 4.

Fig. 8. (a) GMI performance under different SNRs and PS degrees, which have been normalized to each SNR (The value ranges from 0 to 1, where 0 represents the lowest GMI, and 1 represents the highest GMI). (b) The SNR and the MI of the signals without EFNS. (c) The SNR and the MI of the signals with EFNS.

Fig. 9. Comparison between the combined PS and DRE and the joint shaping.