Due to increasingly large computational resources, modern neural networks are severely constrained due to their processing speed and energy consumption. Optical neural networks (ONNs), which use photonic structures to process signals at the physical level as an alternative to the computation in the electronic domain provided by traditional neural networks, are an attractive approach to implementing ultra-high-speed, low-energy parallel computation. Nevertheless, current training processes for electronic domain neural networks are optimized from gradient-based training methods, such as backpropagation, not compatible with ONNs with gradient-free features. In this work, a stochastic function-based gradient-free training method, i.e., stochastic function direct feedback alignment (SF-DFA) is demonstrated and evaluated. SF-DFA trains a gradient-free system using stochastic matrices and functions to replace the weights and gradients of the nodes in neural networks. Thus, it is feasible to train ONNs without a prior knowledge of the photonic system and its gradients. In addition, implementing such training process on optical hardware is also known to be possible. A series of studies have been carried out for a spectral slicing neural network (SS-NN) architecture trained by SF-DFA. The SS-NN system uses bandpass filters embedded in optical fiber micro rings to enable slicing of the optical signal spectrum. Our results demonstrate that the training of ONN using SF-DFA can converge efficiently, with higher processing speed and lower energy consumption compared to back-propagation.
KEYWORDS: Telecommunications, Signal to noise ratio, Fiber optic communications, Optical amplifiers, Digital signal processing, Fiber amplifiers, Modulation, Binary data, Systems modeling
Due to the high transmission capacity, optical fiber systems have been widely applied in the modern telecommunication infrastructure to meet the ever-increasing demand of data traffic. Optical amplifiers have been employed to amplify optical signals and to compensate for the transmission losses. They play a key role in relaying the signals in ultra-wideband optical fiber communication systems. However, the amplified spontaneous emission (ASE) noise will be introduced and will pose constraints on the transmission information rates. The mutual information (MI) and the generalized mutual information (GMI) have been applied to evaluate the information rates in communication systems. In this work, we have investigated the impact of ASE noise on the MI and the GMI, and developed corresponding analyses across different modulation formats. Our work aims to explore the limit and requirements on optical amplifiers in next-generation ultra-wideband optical fiber communication systems.
KEYWORDS: Signal to noise ratio, Telecommunications, Modulation, Optical fibers, Digital signal processing, Systems modeling, Interference (communication), Transmittance, Optical amplifiers, Optical communications
Coherent optical fiber systems can achieve long-distance, large-capacity and high data-rate transmissions. The system performance of communication systems is generally evaluated with regard to the data capacity and the transmission reach. In this work, the performance of multi-channel (up to C-band) Nyquist-spaced coherent optical communication systems has been assessed in terms of achievable information rates, transmission distances and signal-to-noise ratios, considering different influencing factors, such as nonlinearity compensation, signal input power and modulation format. Numerical simulations and enhanced Gaussian noise (EGN) model have been carried out for different modulation formats including quadrature phase shift keying (QPSK), 16-ary quadrature amplitude modulation (16-QAM), 64-QAM and 256-QAM. It is found that in C-band (151-channel) Nyquist-spaced systems, the achievable information rates at the transmission distance of 6000 km are 19.3 Tbit/s for dual-polarization QPSK (DP-QPSK), 30.9 Tbit/s for DP-16QAM, 32.0 Tbit/s for DP64QAM and 32.2 Tbit/s for DP-256QAM, respectively, when electronic dispersion compensation is applied only. Such achievable information rates can be increased up to 38.3 Tbit/s for DP-16QAM, 47.2 Tbit/s for DP-64QAM and 47.8 Tbit/s for DP-256QAM, respectively, when the nonlinearity compensation is employed.
The efficient and accurate evaluation of the transmission performance of high-capacity optical communication systems has always attracted significant research attentions. The enhanced Gaussian noise (EGN) model is considered as an excellent solution to predict the system performance taking into account linear and nonlinear transmission impairments. Since the conventional form of the EGN model is complicated and intractable for a fast computation, the closed-form simplification has been regarded as a direction to significantly reduce the computational complexity. However, the accuracy of such a closed-form EGN model becomes a main concern in the application of ultra-wideband optical communication systems. In this work, we have investigated the accuracy of the closed-form EGN model for ultra-wideband optical fiber communication systems, where the performance of the system using electronic dispersion compensation, multi-channel nonlinearity compensation and full-field nonlinearity compensation has been evaluated in terms of symbol rate, number of channels and signal power. Our work will provide an insight on the application of the EGN model in next-generation ultra-wideband long-haul optical fiber communication networks.
KEYWORDS: Signal to noise ratio, Telecommunications, Digital signal processing, Optical fibers, Distortion, Optical communications, Optical amplifiers, Transmittance, Modulation, Fiber lasers
In digital signal processing (DSP) based coherent optical communication systems, the effect of equalization enhanced phase noise (EEPN) will seriously degrade the transmission performance of high-capacity optical transmission system. In this paper, we have investigated the influence of EEPN on 9-channel 32-Gbaud dual-polarization 64-ary quadrature amplitude modulation (DP-64QAM) Nyquist-spaced superchannel optical field trial by using electronic dispersion compensation (EDC) and multi-channel digital backpropagation (MC-DBP). The deteriorations caused by EEPN on the signal-to-noise-ratio (SNR) and achievable information rates (AIRs) in high-speed optical communication systems have been studied. The system performance versus back-propagated bandwidth under different laser linewidth have also been demonstrated. The SNR penalty due to the distortion of EEPN achieves ~5.11 dB when FF-DBP is implemented, which informs that FF-DBP is more susceptible to EEPN, especially when the LO laser linewidth is larger. The system AIR versus different transmission distance under different EEPN interference using EDC-only and MC-DBP have also been evaluated, which show that there is a trade-off on the selection of lasers and back-propagated bandwidths to achieve a target AIR.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.