We review margins used in optical networks and review a formerly proposed margin taxonomy. For each category of margins, we review techniques that the network designer can use in order to increase the capacity of optical networks, extend their life, and decrease deployment cost (CAPEX) or total cost of ownership over their life duration. Green field (new network deployments) and brown field techniques (used after initial network deployment) are discussed. The technology needed to leverage the margins and achieve the aforementioned gains are also reviewed, along with the associated challenges.

Design of Low-Margin Optical Networks

Networks are complex interacting systems involving cloud operations, core and metro transport, and mobile connectivity all the way to video streaming and similar user applications.With localized and highly engineered operational tools, it is typical of these networks to take days to weeks for any changes, upgrades, or service deployments to take effect. Machine learning, a sub-domain of artificial intelligence, is highly suitable for complex system representation. In this tutorial paper, we review several machine learning concepts tailored to the optical networking industry and discuss algorithm choices, data and model management strategies, and integration into existing network control and management tools. We then describe four networking case studies in detail, covering predictive maintenance, virtual network topology management, capacity optimization, and optical spectral analysis.

/pdf/machine-learning-for-network-automation-overview-uxolqsw1gf.pdf

Machine learning for network automation: overview, architecture, and applications [Invited Tutorial]

Predicting the quality of transmission (QoT) of a lightpath prior to its deployment is a step of capital importance for an optimized design of optical networks. Due to the continuous advances in optical transmission, the number of design parameters available to system engineers (e.g., modulation formats, baud rate, code rate, etc.) is growing dramatically, thus significantly increasing the alternative scenarios for lightpath deployment. As of today, existing (pre-deployment) estimation techniques for light-path QoT belong to two categories: “exact” analytical models estimating physical-layer impairments, which provide accurate results but incur heavy computational requirements, and margined formulas, which are computationally faster but typically introduce high link margins that lead to underutilization of network resources. In this paper, we explore a third option, i.e., machine learning (ML), as ML techniques have already been successfully applied for optimization and performance prediction of complex systems where analytical models are hard to derive and/ or numerical procedures impose high computational burden. We investigate a ML classifier that predicts whether the bit error rate of unestablished lightpaths meets the required system threshold based on traffic volume, desired route, and modulation format. The classifier is trained and tested on synthetic data and its performance is assessed over different network topologies and for various combinations of classification features. Results in terms of classifier accuracy are promising and motivate further investigation over real field data.

/pdf/machine-learning-method-for-quality-of-transmission-xjwrynatqg.pdf

Machine-learning method for quality of transmission prediction of unestablished lightpaths

Dramatic increase in data traffic, future diffusion of bandwidth-hungry 5G connectivity and evolving services like virtual reality and cloud will require the implementation of the open and elastic network paradigm on backbone optical networks. To support such growth, operators are aiming at lowering optical network costs by deploying multi-vendor disaggregated optical networks supporting white boxes while maintaining target performances. To make this effort successful, a vendor-agnostic assessment of optical performance is needed, and operators and vendors are working together to achieve this. In this paper, we present the vision of the Open Optical Packet Transport Physical Simulation Environment (PSE) Group of the Telecom Infra Project consortium, which aims at developing an open source framework for quality-of-transmission (QoT) assessment for design and operation of multi-vendor optical networks. We validate the PSE QoT estimator—the optical link emulator (OLE) based on the Gaussian noise (GN) model for nonlinear fiber propagation—by comparison with experimental results on the mixed-fiber test-bed network at Microsoft labs, operated in the C-band by commercial transceivers from eight different vendors, with propagation distances up to 1945 km. An excellent agreement between OLE predictions requiring milliseconds of computational time and measurements has been observed for all channels. The observed $\pm$ 0.75 dB of accuracy in OLE predictions can be further reduced by including the stimulated Raman scattering in the GN model and with a more accurate capability to estimate power levels and the amount of amplified spontaneous emission noise. Further work will be also required to automate the measurement process of operational parameters from network equipment, especially optical amplifiers, to streamline the overall QoT estimation process.

Multi-Vendor Experimental Validation of an Open Source QoT Estimator for Optical Networks

For reliable and efficient network planning and operation, accurate estimation of Quality of Transmission (QoT) before establishing or reconfiguring the connection is necessary. In optical networks, a design margin is generally included in a QoT estimation tool (Qtool) to account for modeling and parameter inaccuracies, ensuring the acceptable performance. In this article, we use monitoring information from an operating network combined with supervised machine learning (ML) techniques to understand the network conditions. In particular, we model the penalties generated due to i) Erbium Doped Fiber Amplifier (EDFA) gain ripple effect, and ii) filter spectral shape uncertainties at Reconfigurable Optical Add and Drop Multiplexer (ROADM) nodes. Enhancing the Qtool with the proposed ML regression models yields estimates for new or reconfigured connections that account for these two effects, resulting in more accurate QoT estimation and a reduced design margin. We initially propose two supervised ML regression models, implemented with Support Vector Machine Regression (SVMR), to estimate the individual penalties of the two effects and then a combined model. On Deutsche Telekom (DT) network topology with 12 nodes and 40 bidirectional links, we achieve a design margin reduction of ∼1 dB for new connection requests.

/pdf/modeling-edfa-gain-ripple-and-filter-penalties-with-machine-2i0i4ri1pj.pdf

Modeling EDFA Gain Ripple and Filter Penalties With Machine Learning for Accurate QoT Estimation

Wavelength selective switches (WSSs) that are integrated in reconfigurable optical add-drop multiplexers (ROADMs) induce penalties on the optical signal due to tight optical filtering, which increases as several ROADMs are cascaded in a meshed network. In this letter, we propose and analyze possible configurations for the mitigation of these penalties in optical domain using optical wave shaper (WS). Including one WS in every ROADM node allows transmission of 28 and 32 GBd signals, which are QPSK, 8-QAM, or 16-QAM modulated, through a cascade of 32 and 14 WSS filters, respectively. With an average bandwidth of 33 GHz per WSS, an optical signal to noise ratio penalty below 1 dB at BER=1×10-3 is observed.

On the Mitigation of Optical Filtering Penalties Originating From ROADM Cascade

Next-generation elastic optical interfaces will support a wide range of line rates and modulation formats. Such transmission schemes enable doubling of the channel capacity at the expense of a lower reach, whereas a flexible DWDM grid supports the transport of multiple optical signals within a single frequency slot by packing them closely, thus saving spectral resources. The resulting multitude of options on the line side provides network planners the capability to derive the most suitable one for each individual path inside the network. However, planning and operational complexity should not be overlooked, since aspects such as added spectrum fragmentation can hamper the expected network improvements. This paper investigates how to obtain the best compromise between harnessing the main benefits of next-generation optical interfaces, while keeping the complexity of the underlying system at a reasonable level. In particular, we define a novel network spectral efficiency (SE) metric that enables, on one hand, to highlight the relevance of supporting a flexible grid to improve SE and, on the other hand, to demonstrate that the grid granularity can be based on coarser 50 GHz increments without major penalties. The effectiveness of this parameter has been verified by considering two reference network topologies. Finally, the paper also discusses how future optical interface technology developments will shape network design.

Networking aspects for next-generation elastic optical interfaces

The invention refers a method and an arrangement for channel set up in an optical network. An optical signal path is configured for a certain optical channel signal (OC1) of a WDM-signal. This channel signal (OC1) is on-off-modulated by a modulation test signal (MT1) having a predetermined lower frequency and is generating a channel test signal (OT1). This channel test signal (OT1) is combined with other optical channels (OC2-OCn) to the WDM-signal (WS) and transmitted via said path. At a start node (1) or a downstream node (3, 5) a measurement signal (EMI, EM3) is derived from the complete WDM-signal (WS) without wavelength de-multiplexing. The measurement signal (EM1, EM3) is compared with a correlation signal (MC1) and an obtained power level (PC1) is used to adjust the channel power (PC1, PC2, PC3) to achieve predetermined target power values (PC1-PC4) at different power monitoring points (19, 40, 41, 58). The adjustment step is repeated for all downstream nodes (3, 5) and all further channel signals (OC2-OCn).

Method and arrangement for channel set up in an optical wdm-network

The design of a ROADM architecture is depending on many criteria among with network functionality requirements, cost, footprint and scalability. But architectural choices also impact the performance of the optical channels being transported. The motivation of the present paper is to examine the performance impact of exemplary architectures in dependence on modulation formats and network configurations. The impact of ROADM architecture on the transmission performance in different network scenarios is carefully analysed taking into account both the express layer and the add/drop structure. 100G QPSK and 200G 16-QAM are chosen as representative modulation formats as their design requirements are significantly different not only in the express layer but also in the add/drop part. The papers demonstrates that the reach impact of the ROADM architecture can be significant and should be taken into account when optimizing ROADM architectures for each network domain.

Benoit Clouet

Papers

On the Mitigation of Optical Filtering Penalties Originating From ROADM Cascade

Networking aspects for next-generation elastic optical interfaces

Method and arrangement for channel set up in an optical wdm-network

On the Design Trade-Offs of ROADM Architectures