Optical Network Design Research Articles

Ensemble learning of diffractive optical networks

A plethora of research advances have emerged in the fields of optics and photonics that benefit from harnessing the power of machine learning. Specifically, there has been a revival of interest in optical computing hardware due to its potential advantages for machine learning tasks in terms of parallelization, power efficiency and computation speed. Diffractive deep neural networks (D2NNs) form such an optical computing framework that benefits from deep learning-based design of successive diffractive layers to all-optically process information as the input light diffracts through these passive layers. D2NNs have demonstrated success in various tasks, including object classification, the spectral encoding of information, optical pulse shaping and imaging. Here, we substantially improve the inference performance of diffractive optical networks using feature engineering and ensemble learning. After independently training 1252 D2NNs that were diversely engineered with a variety of passive input filters, we applied a pruning algorithm to select an optimized ensemble of D2NNs that collectively improved the image classification accuracy. Through this pruning, we numerically demonstrated that ensembles of N = 14 and N = 30 D2NNs achieve blind testing accuracies of 61.14 ± 0.23% and 62.13 ± 0.05%, respectively, on the classification of CIFAR-10 test images, providing an inference improvement of >16% compared to the average performance of the individual D2NNs within each ensemble. These results constitute the highest inference accuracies achieved to date by any diffractive optical neural network design on the same dataset and might provide a significant leap to extend the application space of diffractive optical image classification and machine vision systems.

油气管道SDH光传输网关键设计

油气管道SDH光传输网关键设计

Optical Cable Backbone Network Design in West Kalimantan Province Using OTN DWDM Technology

In designing the backbone network for the province of West Kalimantan using Dense Wavelength Division Multiplexing (DWDM) and Optical Transport Network (OTN) devices which cover 14 districts and cities consisting of 30 segments, of which 30 segments consist of 14 DWDM OTN sites and 8 Optical Land Amplifier sites ( OLA). From the results of the design and simulation of the backbone network, it was found that the value of the Optical Link Power Budget (OLPB) < 50 dBm where this value is the ideal value in deploying a DWDM OTN backbone network, and the range of the Optical Signal Noise Ratio (OSNR) is between 20 dBm < OSNR < 40 dBm, which is the ideal value for OSNR, and has an Optical Rise Time Budget (ORTB) < 60 ps, and also has a segment margin system of > 7 dB.

REQUIREMENTS FOR OPTICAL TRANSPORT NETWORKS FOR SUCCESSFUL IMPLEMENTATION OF 5G TECHNOLOGY

Background. The amount of data generated, transmitted and accumulated in the world is constantly growing every year. On the other hand, users need new services and information exchange capabilities. 5G technology is able to meet the requirements of subscribers to some extent. Therefore, it is important to consider the main features of fifth generation mobile networks and to determine the requirements for the design of optical transport networks that will ensure the successful implementation of 5G technology.Objective. The aim of the paper is to determine the features of design and characteristics of optical transport networks to implement the capabilities of 5G technology.Methods. The analysis of the publications devoted to modern optical networks and their optical components allowed defining the basic requirements to design of optical transport networks and their characteristics.Results. It is established that optical transport networks should provide increased bandwidth; possibility of mass optical cross-connections in ROADM/OXC nodes that support addition/removal of any wavelength from/in any direction; intelligent operation and maintenance of the network.Conclusions. The defined requirements for the design of optical transport networks and their characteristics will ensure the successful implementation of the capabilities of 5G technology, which will meet the needs of subscribers.

Scale-, Shift-, and Rotation-Invariant Diffractive Optical Networks

Recent research efforts in optical computing have gravitated towards developing optical neural networks that aim to benefit from the processing speed and parallelism of optics/photonics in machine learning applications. Among these endeavors, Diffractive Deep Neural Networks (D2NNs) harness light-matter interaction over a series of trainable surfaces, designed using deep learning, to compute a desired statistical inference task as the light waves propagate from the input plane to the output field-of-view. Although, earlier studies have demonstrated the generalization capability of diffractive optical networks to unseen data, achieving e.g., >98% image classification accuracy for handwritten digits, these previous designs are in general sensitive to the spatial scaling, translation and rotation of the input objects. Here, we demonstrate a new training strategy for diffractive networks that introduces input object translation, rotation and/or scaling during the training phase as uniformly distributed random variables to build resilience in their blind inference performance against such object transformations. This training strategy successfully guides the evolution of the diffractive optical network design towards a solution that is scale-, shift- and rotation-invariant, which is especially important and useful for dynamic machine vision applications in e.g., autonomous cars, in-vivo imaging of biomedical specimen, among others.

Design and analysis of hybrid optical distribution network for worst-case scenario of E2-class symmetric coexistence 80 Gbps TWDM NG-PON2 architecture for FTTX access networks

In this paper, we investigated the cost-effective design and validation of E2-class coexistence 80 Gbps symmetric 8 × 8 Time and Wavelength Division Multiplexing (TWDM) NG-PON2 optical access network considering worst-case scenario for the deployment of Fiber-to-the-X (FTTX) access network using hybrid optical distribution network (ODN). Full-service access network (FASN) unanimously agreed and ITU-T G.989.2 described recommendations for TWDM PON NG-PON2 technology to cater to the need of ever-increasing demand for higher bandwidth and data speed beyond 10 Gbps per user through access networks. Wavelength select (WS) and wavelength-routed (WR) ODN for downstream (D/S) and upstream (U/S) channels symmetric coexistence functionality supporting 80 Gbps accommodating GPON and XGS-PON legacy channels i.e.4-TWDM and 4 point-to-point (PtP) WDM channels operating at 1.25/2.5/10 Gbps respectively in D/S and U/S direction with error free higher splitter configuration of 2304 at 20 km, 640 at 40 km, 576 at 60 km. Precise calculations are done to ensure 11 dBm and 9 dBm ODN launch power (PTODN) to implement worst-case scenario for symmetric E2-class TWDM NG-PON2 network. Calculated inter-channel cross-talk (Cc) is -33.57 dB/-34.74 dB and corresponding cross-talk power penalty (Pc) is 0 dB for D/s and U/S channels respectively. Simulation results demonstrated and verified with the recommendations made in ITU-T G.989.2 confirmed that the proposed network configuration supports extended to reach upto 60 km in D/S and 80 km in U/S direction delivering incremental receiver sensitivity (Rxs) as -34.40/-34.40/42.21 dBm and -34.11/-37.25/-33.37 dBm, ODN path loss 29.40 dB.

Virtualization of elastic optical networks and regenerators with traffic grooming

An elastic optical network (EON) plays an important role in transport technology for virtualization of networks. A key aspect of EONs is to establish lightpaths (virtual links) with exactly the amount of spectrum that is needed and with the possibility of grooming, the process of grouping many small traffic flows into larger units, creating a super-lightpath. Grooming eliminates the need for many guard bands between lightpaths and also saves transceivers; however, it often leads to the need to perform optical–electrical–optical conversions to multiple-data-rate optical signals at intermediate nodes. The aim of this paper is to provide a mixed-integer linear programming (MILP) formulation, as well as heuristic and meta-heuristic approaches, for the design of multiple virtual optical networks (VONs) in an elastic optical substrate network with bandwidth-variable lightpaths, modulation format constraints, and virtual elastic regenerator placement. Traffic grooming is allowed inside each VON, and a distance-adaptive modulation format technique is employed to guarantee efficiency in terms of bandwidth for a physical substrate, subject to several virtual topologies. A reduced MILP formulation without grooming capability is also proposed for comparison. The complete MILP formulation jointly solves the virtual topology design, regenerator placement, and grooming problems, as well as the routing, modulation, and spectrum assignment (RMSA) problem. The reduced MILP formulation, heuristics, and meta-heuristic, on the other hand, separate the virtual topology design problem from the RMSA problem. It is shown that the grooming approach can provide good results, since it solves the problem for a complete design when compared to the approach without grooming. Furthermore, heuristic solutions for large networks are proposed, which present good performance (in terms of saving spectrum) for the design with large instances.

QoS-aware protection in elastic optical networks with distance-adaptive and reconfigurable modulation formats

• Proposal on QoS-aware protection combined with reconfigurable modulation formats. • A novel network design formulation for QoS-aware protection. • Protection based on traffic differentiation brings significant spectral saving. Elastic optical networks have recently emerged as one of the most promising technologies for creating next-generation optical core networks thanks to the adaptive spectrum allocation mechanism. In order to achieve higher spectrum efficiency and exploit elastic optical networking technology, in this paper, we introduce a novel perspective on re-designing dedicated path protection for elastic optical networks. First, inspired by the observation that, in distance-diverse network topologies, the transmission margin between the working and protection route could be wide enough and therefore, the most feasibly spectrum-efficient modulation format could be assigned according to the specific condition of each route rather than applying a single modulation format that is able to cope with the worse condition of the route as currently adopted in traditional designs. Such more granular in transmission adaption is powered by the fast reconfiguration of state-of-the-art elastic transponders and this paves new opportunities in leveraging the spectrum efficiency. Second, the characteristics of traffic traveling in fiber links could be furthermore exploited to tailor the protection services to the actual requirement instead of enforcing the same level of protection for the entire traffic. Specifically, premium traffic today accounts for only 25 % and the remained traffic is best-effort and in taking advantage of this insight, dedicated path protection could be re-designed to ensure just-enough bandwidth provision for premium traffic in case of failures while possibly accepting best-effort traffic discard. In exploiting both the reconfiguration capability and traffic differentiation, a new quality-of-service aware protection framework for the design of elastic optical networks with the capability of distance-adaptive and reconfigurable modulation format assignment is provided. We introduce the node architecture supporting our proposal and provide an optimal design formulation in the form of integer linear programming model to maximize the spectral utilization efficiency. We have also applied extensively experimental simulations to evaluate the performance of our proposed solution in comparison with that of appropriate conventional approaches on a realistic network topology, i.e. COST239. The obtained numerical results prove that our developed solution significantly outperforms the conventional ones, in terms of spectrum efficiency, and it has been revealed that permitting reconfiguration of transmission parameters in switching from working routes to protection ones could result up to 20 % spectral saving in most favorable conditions. Furthermore, incorporating the protection service differentiation could substantially boost spectrum efficiency. For the present ratio of premium traffic– roughly 25 % , our proposal reveals that a spectral saving of up to more than 80 % could be attained and on the whole, the developed solution tends to be more profitable with the medium and high-load traffic conditions.

On the suitability, requisites, and challenges of machine learning [Invited

The introduction of 5G, the increasing number of connected devices, and the exponential growth of services relying on connectivity are pressuring multilayer networks to improve their scaling, efficiency, and controlling capabilities. However, enhancing those features consistently results in a significant amount of complexity in operating the resources available across heterogeneous vendors and technology domains. Thus, multilayer networks should become more intelligent in order to be efficiently managed, maintained, and optimized. In this context, we are witnessing an increasing interest in the adoption of artificial intelligence (AI) and machine learning (ML) in the design and operation of multilayer optical transport networks. This paper provides a brief introduction to key concepts in AI/ML, highlighting the conditions under which the use of ML is justified, on the requisites to deploy a data-driven system, and on the challenges faced when moving toward a production environment. As far as possible, some key concepts are illustrated using two realistic use-cases applied to multilayer optical networks: cognitive service provisioning and quality of transmission estimation.

Seven-channel 1 Gbps TWDM coexistence architecture supporting 65 Gbps optical link for next-generation passive optical network 2–based FTTX access networks

Abstract In this paper, we have presented the design and simulation of a 7-channel next-generation passive optical network (NG-PON2) network for the deployment of Fiber-to-the-X (FTTX) access network. Coexistence architecture is proposed, designed and simulated for the implementation of NG-PON2 access network. In a coexistence architecture approach, legacy PON networks like Gigabit passive optical network (GPON) PON, 10GPON, etc. and wavelength division multiplexing (WDM)-PON supporting point-to-point connectivity are designed and simulated together. A 4 × 4 WDM-PON in which each channel carrying data at 2.5 Gbps data rate is capable of supporting a throughput channel capacity of 65.5 Gbps. NG-PON2 network is designed and simulated at 187.1, 187.2, 187.3 and 187.5 to 187.8 THz wavelengths in downstream direction for different link distances from 40 to 80 km looking into the requirement of reach of access network for future cities. The network performance parameters such as bit error rate (BER), quality factor (Q-factor), signal-to-noise ratio using the Optisystem-16 simulator at above data rates and link distances. Further, channel capacity estimation is done for single-mode fiber channel coexistence NG-PON2 configuration up to 80 km supporting BER e-13 and Q-factor 7 for WDM link and BER e-12 and Q-factor 7 for a legacy network supporting almost-1 Gbps data rate to 65 users and 100 Mbps to 512 user.

A new branch-and-cut approach for the generalized regenerator location problem

In optical networks, a signal can only travel a maximum distance (called optical reach) before its quality deteriorates, needing regenerations by installing regenerators at network nodes. Such an optical reach is an important property of a transmission system, which is a necessary ingredient for enabling optical bypass and thus significantly affects optical network design. In this paper, we study the generalized regenerator location problem (GRLP) where we are given a set S of candidate locations for regenerator placement and a set T of network nodes required to communicate with each other. The GRLP is to find a minimal number of network nodes for regenerator placement, such that for each node pair in T, there exists a path of which no subpath without internal regenerators has a length greater than the given optical reach. Starting with an existing set covering formulation of the problem, we first study the facial structure of the associated polytope. Making use of these polyhedral results, we then present a new branch-and-cut solution approach to solve the GRLP to optimality. With benchmark instances and newly generated instances, we finally evaluate our approach and compare it with an existing method. Computational results demonstrate efficacy of our approach.

Flexible coherent UDWDM-PON with dynamic user allocation based on limited-tunability lasers

Coherent wavelength division multiplexing (WDM) technologies have leveraged the optical communication systems in core networks, increasing the fiber capacity by transmission with advanced modulation formats and mitigation of impairments with digital signal processing. However, these solutions are too expensive for access networks, where cost, power budget, and footprint are limited. Hence, the key technology will be developing low-cost coherent transceivers providing an excellent selectivity and giving high sensitivity, which allows high splitting ratios. This paper reports an experimental design of a low-cost coherent ultra-dense WDM passive optical network (UDWDM-PON) with 6.25 GHz channel spacing. The users’ optical network unit (ONU) is built employing coherent transceivers with two paired low-cost distributed feedback (DFB) lasers, one as the local oscillator and another as the transmitter, offering simplicity and low-cost hardware; likewise, the optical line terminal (OLT) at the central office can profit from the same design. The ONU DFB lasers have wavelengths with limited thermal tunability, controlled by a thermo-electric cooler, which is used to allocate the wavelengths. A medium access control (MAC) at the OLT manages the spectrum channel allocation for ONUs, demanding connection when activation is requested: the OLT furnishes an optical carrier wavelength for the ONU to obtain connection by a control algorithm, assigning a down-channel and another paired up-channel assigned to the ONU DFB transmitter. The MAC can reassign the channels because of interference or collision in a dynamic wavelength allocation. Measures in an activation process and in channel reassignment have been performed in environmental conditions, including control signals and the physical parameters of DFB lasers, demonstrating the practical viability of the PON scaling from 32 up to 256 wavelength channels.

A novel design of energy efficient bidirectional passive optical green networks using reduced buffer with small packet size technique

AbstractEnergy saving and reduction in the emission of carbon dioxide, sulfur, etc. have become one of the prime focus for the deployment of next generation networks. Due to less power consuming, passive optical networks are the best promising solution for such networks. We have proposed different aggressive schemes to reduce the energy consumption viz. no buffer, with buffer, reduced buffer, and reduced buffer with small packet. We evaluated the performance of proposed schemes for downstream and upstream by exploring the relationship between power consumption, delay, and average traffic load. The “reduced buffer with small packet” scheme has shown a reduction of more than 35% of total leakage power at the central office with the compromise of delay in packets receiving.

An overview of ML-based applications for next generation optical networks

Over the past few decades, the demand for the capacity and reliability of optical networks has continued to grow. In the meantime, optical networks with larger knowledge scales have become sources of numerous heterogeneous data. In order to handle these new challenges, many issues need to be resolved, among which the low-margin optical networks design, power optimization, routing and wavelength assignment (RWA), failure management are quite important. However, the use of traditional algorithms in the above four applications shows some shortcomings. Fortunately, artificial intelligence (AI), especially machine learning (ML), is regarded as one of the most promising methods to overcome these shortcomings. In this study, we review the applications of ML methods in solving these four issues. Although many ML-based researches have emerged, the applications of ML techniques in optical networks still face challenges. Therefore, we also discuss some possible future directions of investigating ML-based approaches in optical networks.

Inverse design of an integrated-nanophotonics optical neural network

Artificial neural networks have dramatically improved the performance of many machine-learning applications such as image recognition and natural language processing. However, the electronic hardware implementations of the above-mentioned tasks are facing performance ceiling because Moore’s Law is slowing down. In this article, we propose an optical neural network architecture based on optical scattering units to implement deep learning tasks with fast speed, low power consumption and small footprint. The optical scattering units allow light to scatter back and forward within a small region and can be optimized through an inverse design method. The optical scattering units can implement high-precision stochastic matrix multiplication with mean squared error <10-4 and a mere 4 × 4 μm2 footprint. Furthermore, an optical neural network framework based on optical scattering units is constructed by introducing “Kernel Matrix”, which can achieve 97.1% accuracy on the classic image classification dataset MNIST.

Design of Optical Tunnel Switching Networks for Big Data Applications

In this paper, we proposed large-scale optical tunnel switching networks based on the Torus topology network with WSS (Wavelength Selective Switch) for future big data applications. All nodes of the large-scale optical tunnel switching networks use WSS switch modules, and the communications between nodes use multiple λs (wavelengths), where a tunnel is established with a wavelength which can be reused. The widely used MEMS (Micro-Electro-Mechanical Systems) and LCoS (Liquid Crystal on Silicon) technologies are all millisecond-level switching speeds, so the frame size of the optical frame switch is very large, and this will reduce switching performance. Therefore, they are only suitable for optical tunnel switching networks design, but are not suitable for optical frame switch design. This multi-plane Torus topology network architecture not only increases network throughput, but also has fault tolerance to increase network reliability. When the traffic is changed, the number of tunnels between nodes can be scheduled in time to balance the load traffic and avoid traffic loss. Therefore, it can not only schedule the number of tunnels in time to balance the load traffic, in order to avoid traffic loss, but also because the channel is fixedly established, this does not generate any buffer delay, and this because of the transmission using optical transmission unlimited speed, so it is a good choice for future big data applications that require high speed, high bandwidth and low latency.

Spectrum Utilization with Increasing Traffic, Based on ILP Model in Elastic Optical Networks

The traffic volume due to bandwidth hungry multimedia and internet applications is growing exponentially and demands a significant change in communication networks. Due to the high speed communication requirements present researchers are giving interest in efficient, adaptive and cost effective optical network design. Recent works in optical networking using optical fiber communication focusing on the efficient utilization of resources by flexibly utilizing the spectrum and adaptive networking to support a better quality of service (QoS). The flexible grid in optical spectrum and adaptive optical networking save spectrum with improvement in energy and spectral efficiency. Simple modulation of a signal can support lower bit rate transmissions, which are more sophisticated for higher bit rates. Recent advances on hardware such as optical transponders and switches make it more feasible to develop a new elastic-optical networking (EON) paradigm. Current research efforts are on new innovations of optical transponders and optical switches to support EON functionalities. The measure of spectrum utilization factor with respect to the value of traffic requirement increases the flexibility of the networks. The Integer Linear Programming (ILP) is used to extract the optimal solution for analysis of utilized frequency slots with available slots and increased traffic in the network. This framework also calculates the spectrum utilization for different traffic in order to choose required frequency slot for the EON to provide uninterrupted network service.

Design of task-specific optical systems using broadband diffractive neural networks

Deep learning has been transformative in many fields, motivating the emergence of various optical computing architectures. Diffractive optical network is a recently introduced optical computing framework that merges wave optics with deep-learning methods to design optical neural networks. Diffraction-based all-optical object recognition systems, designed through this framework and fabricated by 3D printing, have been reported to recognize hand-written digits and fashion products, demonstrating all-optical inference and generalization to sub-classes of data. These previous diffractive approaches employed monochromatic coherent light as the illumination source. Here, we report a broadband diffractive optical neural network design that simultaneously processes a continuum of wavelengths generated by a temporally incoherent broadband source to all-optically perform a specific task learned using deep learning. We experimentally validated the success of this broadband diffractive neural network architecture by designing, fabricating and testing seven different multi-layer, diffractive optical systems that transform the optical wavefront generated by a broadband THz pulse to realize (1) a series of tuneable, single-passband and dual-passband spectral filters and (2) spatially controlled wavelength de-multiplexing. Merging the native or engineered dispersion of various material systems with a deep-learning-based design strategy, broadband diffractive neural networks help us engineer the light–matter interaction in 3D, diverging from intuitive and analytical design methods to create task-specific optical components that can all-optically perform deterministic tasks or statistical inference for optical machine learning.

Network coding for improving throughput in WDM optical networks with dedicated protection

Network coding (NC) has been recently emerging as an excellent technique for leveraging network efficiency. In optical networks, NC and protection appear to be a good match and indeed, NC could be effectively exploited for greater protection efficiency thanks to the opportunity of combining signals rather than duplication in the traditional approach. Motivated by this observation, we investigate the application of a simple and practical NC technique based on XOR operation to improve the network throughput in survivable optical networks. To this end, we propose the NC-based protection whose speed recovery is on a par with the conventional $$1+1$$ dedicated protection while the capacity usage could be more efficient, paving the way for improving throughput in constrained capacity conditions. A novel mixed integer linear programming model is developed for survivable optical networks design employing the proposed NC-based protection. The model attempts to optimally determine the demands to serve, the routing for served demands and NC assignment among demands aiming at maximizing the network throughput in the constraint of bandwidth limitation. Extensive numerical evaluations on realistic topologies are provided to highlight the advantages of our NC-based protection proposal compared with the conventional dedicated protection in improving the network throughput. Encouraging results are observed when the new protection proposal based on NC could accommodate up to $$30\%$$ more traffic than the conventional case.

On the spectrum-efficiency of QoS-aware protection in elastic optical networks

Elastic optical networks technologies feature several benefits including higher spectral efficiency, system capacity and therefore have been widely viewed as the promising paradigm for next-generation optical transport networks. Dedicated path protection for elastic optical networks has been of ever-increasing importance in the era of information society thanks to its near-immediate recovery speed and remarkably simple operations in guaranteeing the connectivity of information flows. Inspired by the observation that the today premium network traffic traveling in fiber links accounts for roughly 25% of total traffic and the best-effort traffic constitutes the remaining, we propose a new design for elastic optical networks with dedicated path protection tailoring to the actual traffic profile rather than enforcing the same protection service for both traffic types as with conventional approaches. Our proposal exploits the flexible baud-rate transponders shared between the working and protection mode, featuring lower baud-rate operation for protection mode to provide just enough bandwidth for premium traffic. To maximize the impact of protection service differentiation, we present a novel integer linear programming taking into account the proposed quality-of-service-aware protection aiming at minimizing the spectrum usage. Numerical results on a realistic topology, COST239, are provided to assess the efficiency of our proposal compared to the traditional approach and it has been highlighted that having quality-of-service-aware protection being adapted to the actual traffic characteristic presents significant spectrum savings. With the current estimate of premium traffic of 25%, our study suggests that more than 30% spectral saving could be achieved and generally the low-load condition is more profitable than the medium and high-load ones.