We propose a novel low complexity two-stage frequency offset estimation (FOE) scheme based on modified Zoom-FFT for square 16/64-quadrature amplitude modulation (QAM) coherent optical systems. The first stage roughly estimates the frequency offset through a small size FFT. After frequency shifting, the modified Zoom-FFT can select and reconstruct the spectrum of interest by time domain average. Since the average provides a good estimate of the amplitudes of the frequency components, we can accurately find the peak after the second-stage FFT. We conduct simulations and compare the normalized frequency variance of the proposed scheme with different algorithms, e.g., the single-stage FFT and the Chirp-Z transform (CZT). The simulation results show that the proposed FOE scheme has the similar normalized frequency variance as other algorithms of $4.0\times 10^{-9}$ and $1.6\times 10^{-9}$ for 16QAM and 64QAM, respectively. The numerical simulations and experimental results demonstrate that the proposed scheme can achieve comparable performance, while the complexity is reduced by 86.6% (81.2%) and 60.9% (61.2%) for 16QAM (64QAM) compared with single-stage FFT and FFT+CZT, respectively.

Transfer learning (TL) has been demonstrated its feasibility on fast remodeling for fiber nonlinearity equalization. It will be very efficient with fine-tuning rather than retraining when the channel parameters have been changed, e.g. the optical launch power and fiber distance. However, the performance is dependent on the highly correlation between the source and target tasks. An improper source domain cannot improve the efficiency or even negative transfer. In this paper, we propose a meta-learning assisted source domain optimization scheme in single- and multi-channel transmissions. We construct the support dataset and query dataset from the information of the inner-task and the across-task to train the neural network so that we can find a unified source domain and efficiently adapt to different tasks with fewer epochs . We conduct an experiment to transmit 120-Gb/s dual polarization (DP) 16-QAM signal over 800-km standard single mode fiber (SSMF). The required epochs for convergence can be significantly reduced and there is slight Q-factor improvement with meta-learning algorithm compared with retraining and conventional TL. We also analyze the TL strategy in multi-channel system. The simulation results show that meta-learning can also help to optimize the source domain with lower complexity.

We propose and experimentally demonstrate a simplified multiple-input multiple-output Volterra filter (MIMO-VF) in a polarization division multiplexed (PDM) twin single side-band (SSB) scheme based on two offset carriers. The right side-band (RSB) and left side-band (LSB) signals of the twin-SSB in each polarization are separated by an optical filter, and the crosstalk between the RSB and LSB is mitigated in the receiver digital signal processing (DSP) by the proposed simplified MIMO-VF, which is combined with one two-stage equalizer, i.e., a linear 2×2 butterfly MIMO equalizer and two independent VFs. With the proposed simplified MIMO-VF, we experimentally demonstrate a 400-Gb/s four-level pulse amplitude modulation (PAM-4) signal transmission over an 80-km standard single mode fiber (SSMF). Compared with the traditional MIMO-VF, the proposed simplified MIMO-VF can retain the same transmission performance and reduce the computational complexity by approximately 57.9%.

In the canonical digital back-propagation (CDBP) of compensating for optical fiber nonlinearity, the aim is to invert the nonlinear Schrödinger equation (NSE). Therefore, the virtual link mirrors the fiber link for all the deterministic parameters. In theory, it can eliminate the deterministic linear and nonlinear impairments. However, the CDBP inherently does not consider the random noise along the fiber link. Meanwhile, the ultimate aim of optical receivers is to lower the BER or increase the SNR. To this aim, the CDBP may become sub-optimum in the presence of random noise. In this paper, we explicitly prove that the performance of CDBP can be improved by simply adjusting the power map in the virtual link. We call it imbalanced DBP (iDBP), since the parameters in the virtual link do not mirror those in the fiber link. Then by tuning the signal power in the virtual link, we derive the closed-form expressions of iDBP for single-span transmissions, and show that the SNR and information capacity can be increased, compared with CDBP. For multi-span transmissions, we demonstrate the improved performance of iDBP by simulation.

We thoughtfully investigate the feasibility of reducing the mode-coupling caused crosstalk by optimizing the practical parameters of the few-mode fiber (FMF), based on the linear coupling theory. According to our simulation results, the low crosstalk FMF could be achieved with the small fiber core, the low refractive-index (RI) related parameters, the long operating-wavelength and the low core-eccentricity (CE). Moreover, the transmission behaviors over the fixed coupling and the random coupling scenarios are also discussed. The power distribution and the signal to interference plus noise ratio (SINR) evolutions along FMF reveal the impacts from the mode-coupling induced crosstalk in the spatial division multiplexing (SDM) transmission system. The random nature of the mode coupling suggesting the turbulent results are obtained in our simulation, confirmed the observation in the SDM test. And only a 1.23 dB SINR degradation of 16-QAM signals can be realized after 100 km transmission with the optimized CE of lower than 4 × 10 ⁻³ .

As the explosive growth of global IP traffic, the bandwidth growth of optoelectronic devices in intensity modulation/direct detection (IM/DD) systems has begun lagging the capacity increasing of intra-datacenter interconnects (DCIs). The partial response (PR) signaling and various advanced equalizers have been introduced to solve the band-limited effect. In this paper, we propose and experimentally demonstrate a joint PR equalization (PRE) with mapping-rule-based simplified maximum likelihood sequence estimation (MRS-MLSE) scheme in a 256-Gb/s 4-ary pulse amplitude modulation (PAM-4) transmission system. The MRS-MLSE compresses the candidate alphabet of each PRE symbol according to the mapping rule of PRE signaling, resulting in a 41.2% computational complexity reduction compared to the traditional full-state MLSE. Moreover, we design a “Monitor-Activate-Correct” structure that activates the MRS-MLSE to correct errors when detecting error propagations caused by the fundamental 1/(1+D) decoder. The experimental results show that the proposed method can suppress the error propagation in the 1/(1+D) decoder effectively, and achieves a similar performance as the full-state MLSE decoder, which has only 0.3-dB receiver sensitivity penalty compared to it. The statistic results of various transmission scenes indicate that the MRS-MLSE is activated to process lower than 3.74% of total symbols at the KP4-FEC threshold of 2.4×10 ^-4 .