Optical performance monitoring using digital coherent receivers and convolutional neural networks

Generalized optical signal-to-noise ratio monitoring using a convolutional neural network for digital coherent receivers

In this Letter, we propose a generalized optical signal-to-noise ratio (GOSNR) monitoring scheme using a convolutional neural network trained on constellation density features acquired from a back-to-back setup and demonstrate accurate GOSNR estimations for links having different nonlinearities. The experiments were carried over dense wavelength division multiplexing links configured on 32-Gbaud polarization division multiplexed 16-quadrature amplitude modulation (QAM) and demonstrated that the GOSNRs are estimated within 0.1 dB mean absolute error with maximum estimation errors below 0.5 dB on metro class links. The proposed technique does not require any information about the noise floor in the conventional spectrum-based means and therefore is readily deployable for real-time monitoring.

Blind and low-complexity modulation format identification based on signal envelope flatness for autonomous digital coherent receivers

Modulation format identification (MFI) is a critical technology for autonomous digital coherent receivers in next-generation elastic optical networks. A novel and simple MFI scheme, to the best of our knowledge, based on signal envelope flatness is proposed without requiring any training or other prior information. After amplitude normalization and partition, the incoming polarization division multiplexed (PDM) signals can be classified into quadrature phase shift keying (QPSK), 8 quadrature amplitude modulation (QAM), 16QAM, and 64QAM signals according to envelope flatnesses R₁, R₂, and R₃ of signals in different amplitude ranges. The feasibility of the proposed MFI scheme is first verified via numerical simulations with 28 GBaud PDM-QPSK/-8QAM/-16QAM/-64QAM signals. Only by using 4000 symbols can the proposed MFI scheme achieve a 100% correct identification rate for the four modulation formats over a wide optical signal-to-noise ratio (OSNR) range. Proof-of-concept experiments among 28 GBaud PDM-QPSK/-8QAM/-16QAM systems under back-to-back and long-haul fiber transmission links are implemented to further demonstrate the effectiveness of the proposed MFI scheme. The experimental results show that the proposed MFI scheme can obtain a 100% correct identification rate when the OSNR value of each modulation format is higher than the threshold corresponding to 7% FEC and is resilient towards fiber nonlinearities. More importantly, the proposed MFI scheme can significantly reduce computational complexity.

Blind Modulation Format Identification Based on Principal Component Analysis and Singular Value Decomposition

As optical networks evolve towards flexibility and heterogeneity, various modulation formats are used to match different bandwidth requirements and channel conditions. For correct reception and efficient compensation, modulation format identification (MFI) becomes a critical issue. Thus, a novel blind MFI method based on principal component analysis (PCA) and singular value decomposition (SVD) is proposed. Based on square operation and PCA, the influence of phase rotation is removed, which avoids phase rotation-related discussions and training. By performing SVD on the density matrix about constellation, a denoise method is implemented and the quality of the constellation is improved. In the subsequent processing, the denoised density matrix is used as the feature of the support vector machine (SVM), and the identification of seven modulation formats such as BPSK, QPSK, 8PSK, 8QAM, 16QAM, 32QAM and 64QAM is realized. The results show that lower OSNR values are required for the 100% accurate identification of all modulation formats to be achieved, which are 5 dB, 7 dB, 8 dB, 11 dB, 14 dB, 14 dB and 15 dB. Moreover, the proposed method still retains the advantage, even when the number of samples decrease, which is beneficial for low-complexity implementation.

Constellation-based identification of linear and nonlinear OSNR using machine learning: a study of link-agnostic performance

We demonstrate accurate estimation of generalized optical signal to noise ratio (GOSNR) for wavelength division multiplexed fiber communication systems using an experimentally trained multi-tasking convolutional neural network while simultaneously estimating linear and nonlinear noise contributions. Using dual-polarized 32-GBaud 16QAM DWDM links we extract learnable features from constellation density matrices and accurately estimate GOSNR while simultaneously estimating linear and nonlinear contributions. Estimation of the OSNR_ASE, OSNR_NL and GOSNR are demonstrated with < 0.5 dB mean absolute error. We also assess the universality of our model within the regime of metro networks by cross-training with data from such links comprised of different fiber types. We demonstrate a path to a practical universal training method that includes additional link parameters. The methods do not require contiguous high-speed sampling, additional hardware nor transmission of special symbols or patterns and are readily implemented in deployed systems.

Machine Learning Applications for Short Reach Optical Communication

With the rapid development of optical communication systems, more advanced techniques conventionally used in long-haul transmissions have gradually entered systems covering shorter distances below 100 km, where higher-speed connections are required in various applications, such as the optical access networks, inter- and intra-data center interconnects, mobile fronthaul, and in-building and indoor communications. One of the techniques that has attracted intensive interests in short-reach optical communications is machine learning (ML). Due to its robust problem-solving, decision-making, and pattern recognition capabilities, ML techniques have become an essential solution for many challenging aspects. In particular, taking advantage of their high accuracy, adaptability, and implementation efficiency, ML has been widely studied in short-reach optical communications for optical performance monitoring (OPM), modulation format identification (MFI), signal processing and in-building/indoor optical wireless communications. Compared with long-reach communications, the ML techniques used in short-reach communications have more stringent complexity and cost requirements, and also need to be more sensitive. In this paper, a comprehensive review of various ML methods and their applications in short-reach optical communications are presented and discussed, focusing on existing and potential advantages, limitations and prospective trends.

Blind modulation format identification based on improved PSO clustering in a 2D Stokes plane

Blind modulation format identification (MFI) is indispensable for correct signal demodulation and optical performance monitoring in future elastic optical networks (EON). Existing MFI schemes based on a clustering algorithm in Stokes space have gained good performance, while only limited types of modulation formats could be correctly identified, and the complexities are relatively high. In this work, we have proposed an MFI scheme with a low computational complexity, which combines an improved particle swarm optimization (I-PSO) clustering algorithm with a 2D Stokes plane. The main idea of I-PSO is to add a new field of view on each particle and limit each particle to only communicate with its neighbor particles, so as to realize the correct judgment of the number of multiple clusters (local extrema) on the density images of the s₂-s₃ plane. The effectiveness has been verified by 28 GBaud polarization division multiplexing (PDM)-BPSK/PDM-QPSK/PDM-8QAM/PDM-16QAM/PDM-32QAM/PDM-64QAM simulation EON systems and 28 GBaud PDM-QPSK/PDM-8QAM/PDM-16QAM/PDM-32QAM proof-of-concept transmission experiments. The results show that, using this MFI scheme, the minimum optical signal-to-noise ratio (OSNR) values to achieve 100% MFI success rate are all equal to or lower than those of the corresponding 7% forward error correction (FEC) thresholds. At the same time, the MFI scheme also obtains good tolerance to residual chromatic dispersion and differential group delay. Besides that, the proposed scheme achieves 100% MFI success rate within a maximum launch power range of -2∼+6 dBm. More importantly, its computational complexity can be denoted as O(N).