We present an analytical model to estimate the performance penalty due to filtering in coherent systems using finite-length equalizers, reporting a high model accuracy against numerical simulations with a maximum error of 0.1 dB.
This paper presents an orchestration and control architecture for multi-domain optical networks. The proposed solution improves performance predictability by leveraging a modular, closed-loop SDN control plane that enables dynamic service provisioning, optimal path selection, and amplifier control. A key innovation is the integration of a time-varying QoT Digital Twin, which exploits live telemetry to enhance real-time decision-making and network adaptability.
Network automation is essential for accelerating service deployment and optimizing operations, especially in multi-vendor optical transport networks. However, unique interoper-ability challenges arising from vendor-specific implementations require standardized frameworks such as those provided by OpenROADM MSA. This work introduces the Optical Network Robotic Automation Platform (ON-RAP), a fully automated, vendor-agnostic solution that integrates system-level automation with robotic process automation, streamlining OpenROADM-compliant network operations and enabling comprehensive test automation. ON-RAP automates compliance validation, service provisioning, troubleshooting, and multi-vendor simulation, achieving precise configuration, real-time monitoring, and time-efficient fault resolution. ON-RAP enables seamless validation across simulated and real environments, enhancing test efficiency and scalability for network devices and controllers. In addition, it automates report generation to facilitate structured analysis and reduce manual effort.
This paper presents a proactive live migration of the gNodeB Central Unit User Plane (gNB-CU-UP) container in a 5G virtualized Radio Access Network (vRAN) testbed across a multi-vendor OpenROADM optical transport network (OTN). An automated mechanism monitors the CPU usage and dynamically reconfigures midhaul and backhaul wavelength services to maintain low-latency transport. Leveraging Pacemaker and Corosync for seamless Virtual IP (VIP) handover, the proposed mechanism achieves a UE Service Recovery Time of 1.04 sec ensuring minimal disruption.
This paper presents an experimental evaluation of the impact of optical switching equipment on the integrity of sensitive quantum signals, which are fundamental components of quantum networks. Focusing on switching in a Quantum Key Distribution (QKD) network as one example of a quantum network scenario, this study highlights how the impairments introduced by optical switches can significantly degrade the quality of quantum signals, resulting in a noticeable increase in the Quantum Bit Error Rate (QBER). This signal degradation poses serious challenges to the reliability and efficiency of the quantum networks. However, by accurately modeling the effects of switching-induced impairments, designing and implementing an ideal digital twin becomes feasible, which can cut QBER by up to 90% by identifying and addressing root causes and enabling proactive measures. Hybrid quantum-classical networks can benefit from such digital twins through improved quality and scalability.
This paper focuses on the application of an analytical model for faster and reliable transmission characterizations of novel optical networks employing Digital Subcarrier Multiplexing transceivers. This transmission model is thought to be applied in metro-access networks, for which it is first demonstrated for single links. In this work, the model is applied to analyze the system-level performance of a link, including, as a device under test, a 2-Ring Assisted Mach-Zehnder Interferometer (RAMZI) interlacer filter used to perform optical routing at the subcarrier level, and to quickly identify the impact of theparameters of the device in the overall transmission.
Contrary to urban areas, optical fiber deployment in rural scenarios is still in early stage due to high investment and low density of customers. Network operators face a competitive environment in the access segment, where they must deliver fast and reliable Internet access, while offering attractive prices to their users. This work analyzes the energy cost of Fiber to the Building (FttB) networks in rural areas planned using Balanced Passive Optical Network (PON) (BaPON), using a fixed symmetric optical splitter such as 1:16/32/64, etc; and a combination of BaPON and Unbalanced PON (UBaPON) with the use of optical taps in a daisy-chain manner, called Hybrid PON (HyPON) architecture. The analysis is based on the Gigabit Passive Optical Network (GPON) and 10-Gigabit-Symmetrical Passive Optical Network (XGS-PON) standards and four rural scenarios in Germany. The annual power consumption of the network designs are computed and the Operational Expenditures (OpEx) of operating the networks for a period of 20 years is estimated. According to the results, HyPON offers total cost savings of up to 25.32 % in the greenfield scenario of the most sparse region, considering continuous active operation, and serves as a potential candidate for rural PON connectivity.
This study investigates the impact that Point-to-Multipoint (P2MP), traffic growth, and optical aggregation have on the performance of the optical systems and on the necessary hardware required to ensure connectivity across the full network. We leverage the examples of realistic topologies provided by Telecom Italia (TIM) to generalize the network characteristics and allow for statistical analysis of horseshoe-based metro-aggregation scenarios. By comparing the necessary hardware (i.e., number of optical transceivers) in Point-to-Point (P2P) and in P2MP operation, our investigation reveals the extent to which Digital Subcarrier Multiplexing (DSCM) in combination with optical aggregation allows to reduce the number of pluggable units that need to be deployed. In the simulations, under the assumption of a pay-as-you-grow provisioning strategy, we observe a reduction in transceivers in the 33%-45% range depending on the aggregation level and traffic volume after 10 years of operation. In addition to this, we are able to calculate the maximum required cost of the DSCM-capable transceivers with respect to more traditional P2P solutions to produce cost-effective deployments. These results, presented as a function of traffic volume, level of optical aggregation, and relative cost with respect to P2P transceivers, provide an understanding of the overall cost of acquiring DSCM-based pluggable units, and demonstrate that, even though the price point might be higher, the reduction in terms of hardware due to adopting a P2MP approach compensates this aspect and can still allow for the deployment to be more cost-effective.
Over the last two decades, Optical Transport Network (OTN) technology has been a crucial enabler of the core network by efficiently switching and multiplexing traffic with quality-of-service guarantees. Despite their high costs, core network architectures based on traditional monolithic OTN solutions still dominate due to their robustness and performance. Nevertheless, with the increase in high-speed services and the adoption of 400 Gb/s interfaces, novel architectures are needed to fully exploit the transmission capacity of the underlying wavelength division multiplexed (WDM) optical system. In this regard, this paper conducts a techno-economic study that investigates two emerging architectures: disaggregated OTN-over-WDM and Private Line Emulation (PLE)-based IP-over-WDM. They differ in the mechanisms applied to serve packet and circuit-switched services over a common WDM network. Disaggregated OTN-over-WDM replaces monolithic OTN fabrics with white-box Ethernet devices that interwork with IP routers and WDM optical switches. In contrast, PLE-based IP-over-WDM emulates private lines to enable circuit-switched services over IP infrastructures. Both architectures are compared against the standard monolithic OTN-over-WDM solution. Compared to this baseline solution, the results show that disaggregated OTN-over-WDM can achieve cost savings of up to 39%, while the PLE-based architecture can incur cost increments of up to 113%.
Meta operates one of the largest global optical backbone network to support its business across social media, AI training, virtual/augmented reality, etc. The rapid traffic growth on the network leads to significant network design challenges in how to support this growth with limited space/power/cost budget. To address these challenges, we proposed a set of strategies to evolve our backbone network towards a point-to-point architecture, also implementing ZR optics to achieve high efficiency gains in fiber and site investments.
We present a comprehensive experimental analysis of filtering penalties in metro-access scenarios, characterized by cascades of ROADMs and optical amplifiers. This analysis evaluates the impact of various configurations, considering filter bandwidths, data rates, and received optical power. The novelty of this work lies in the derivation and validation of a transceiver model that encompasses filtering penalties, along with the application of this model to facilitate end-to-end lightpath deployment in converged network-as-a-service environments for fog computing, aiming to enhance spectral efficiency.
We present a comprehensive framework for evaluating distributed AI training performance across single and multi-datacenter environments. Our approach combines 2 existing simulators, NeuronaBox's high-fidelity traffic generation and m3's efficient flow-level simulation to analyze various parallelization strategies and synchronization methods much faster than with existing packet-level simulators. We propose and evaluate two datacenter architectures: a single datacenter network (S-DCN) supporting 128k servers across 32 AI pods with non-blocking connectivity, and a distributed datacenter network (D-DCN). Through extensive evaluation, we demonstrate that 800G DCI links can reduce training time by 52.8% compared to 200G links at 50km distance and that introducing an optical circuit switch within a DC achieves superior failure recovery, maintaining flow completion time within 21% of normal operation during pod failures compared to 26% degradation in traditional electronic packet-based architectures, translating to a 2.5 day advantage in recovery scenarios.
We investigate the scaling potential and compute critical performance metrics of a three-tiered hierarchical optical node architecture. The optical node incorporates novel Photonic Integrated Circuit (PIC)-based WaveBand Selective Switches (WBSSs) to implement flexibly-defined band switching, across spatial lanes and degrees of connectivity, ranging from entire optical fibers, flexible-defined bands to individual wavelengths. We also develop a network simulator considering node's multiple switching layers, multiple/portion of bands and all key optical transmission parameters. Simulations reveal high optical Signal to Noise Ratio (OSNR) and low Bit Error Rate (BER) values, particularly for the Full Fiber Switching (FFS) scenario. Moreover, we analyze trade-offs among scalability, component number and complexity, and throughput across various configurations and traffic scenarios. Our results confirm high levels architecture adaptability and efficiency in addressing the evolving demands of future optical networks.
Optical Networking has already played a significant role in the realization of the 5G vision providing increased transport network capacity and low latency for the interconnection of disaggregated radio access and core network elements. However, the evolution towards 6G introducing technologies such as Cell Free-MIMO, Integrated Communication and Sensing and extensive adoption of AI enabled automation brings the need for new features such as further increase in transport network capacity, enhanced granularity in resource allocation as well as configuration flexibility and automation. In this context, optical networking able to effectively support all these requirements can play a key role. This tutorial will provide an overview of 5G and 6G architectural structures and will concentrate on how Optical Networking can support current and upcoming transport network requirements in these environments taking a multilayer approach.
With the expansion of low Earth orbit satellite constellations, optical satellite networks are playing an increasingly important role. One significant issue in optical satellite networks is the frequent laser link failures and corresponding service interruptions. To reduce the possibility of service interruption that is caused by inter-satellite laser link failures, we propose a spatial-temporal shared risk link group (ST-SRLG) model that considers simultaneous link failure risks in both spatial and temporal dimensions. Based on the ST-SRLG model, a dedicated path protection algorithm is proposed to provision backup path that can still survive in case of primary path failures. Simulation results demonstrate that this algorithm can reduce the service interruption rate by 74.03%, thereby enhancing the overall reliability and stability of optical satellite networks.
The design of an optical network's physical topology determines its performance through its graph properties. However, the intelligent and automated design of scalable optical networks remains challenging due to the computational complexity of traditional optimisation methods, deterministic graph generators, and the lack of proper training data. In this paper, we introduce Topology Architect, the first generative AI model for optical network design, capable of generating core topologies from only node counts and geographic coordinates. It is an unsupervised learning framework trained on real networks from our dataset, Topology Bench. The model achieves up to 95% graph similarity, as measured by Wasserstein distances to real networks, and generates user-defined topologies in less than a second. Topology Architect captures the data variability in the latent space 20 times better than a graph's mathematical properties alone. It implicitly integrates multi-objective design principles, scales across different sizes, and presents a novel framework for optical network topology generation.
Quantum Key Distribution Networks (QKDNs) rely on strategic trusted node deployment to overcome distance constraints and enable large-scale quantum-secured communication. We introduce OptiNode, a novel algorithm that determines, using Machine Learning (ML) techniques, the optimal placement of trusted nodes within QKDNs. Leveraging an analytical framework and Bayesian-optimized k-means clustering, OptiNode balances operational feasibility with communicational efficiency using only fixed node positions, which are to be connected, and their expected secure traffic as inputs. An empirical evaluation confirms the algorithm's effectiveness in generating deployable topologies. Our work paves the way for building efficient and high performance QKDN.
Hollow-core fibers (HCF) technologies are evolving rapidly and becoming a candidate for next generation deployable optical fibers. This type of fibers has several advantages compared with traditional ones, e.g., reduced latency, very low nonlinearities, and theoretical lower loss profile. Conversely, as they are sill under development, some uncertainties are present, like their actual loss profile and costs of fabrication/deployment. Moreover, exploiting their potential for improved reach/capacity requires optical amplifiers with higher maximum total output power. Interestingly, HCF proprieties could be leveraged to simplify the optical network infrastructure by reducing the number of in-line amplifiers. This work assesses the potential of HCF to reduce the total number of in-line amplifiers, without compromising end-to-end performance and considering the loss coefficient uncertainty, in reference optical transport networks. Simulation results provide evidence that significant savings in in-line amplifiers are possible, while also guaranteeing high traffic load supported and reduced number of interfaces required, provided that high-power amplifiers and HCF with a loss coefficient equal or lower than traditional silica-based fibers are commercially available.
Advanced optical network technologies, proposed to fulfil the projected demands of future communication networks, are riddled with numerous challenges. Conventional methods for addressing these issues often face scalability limitations in disaggregated systems, prompting a growing reliance on data-driven approaches. Nevertheless, accurately modeling physical phenomena, such as stimulated Raman scattering (SRS), remains difficult since conventional physics-based models can falter due to variations in transmission spans. Machine learning-based models are promising in this regard, but they require huge datasets for enhanced versatility. Another alternative approach is transfer learning (TL), which minimizes the need for extensive datasets by utilizing pre-trained models that incorporate relevant domain knowledge. In this paper, we design and evaluate transfer learning models based on convolutional neural networks (CNN). We use these models to predict the Raman tilt spectra across four fiber configurations in two different testbeds. We achieve less than 0.162 dB mean absolute error which is largely limited by the measurement accuracy.
Optical networks offer an ultra-high transmission capacity and serve various online applications (e.g., 5G, IoT, AR/VR, telemedicine). Preventing faults that cause packet losses or even link interruption becomes vital to ensure the reliability of these networks and, consequently, access to vital online services. Moreover, as the volume of telemetry data rapidly increases, data processing is often done in the cloud, which can open up breaches of unauthorized data access and raise concerns about scalability. Therefore, this work proposes a decentralized federated learning (FL)-based approach that exploits the principal component analysis (PCA) to perform confidentiality-preserving fault detection in optical networks. Unlike centralized FL-based approaches, the PCA is split into several local PCAs trained with subsets of the entire telemetry dataset. Thereafter, each local model exchanges its parameters in a peer-to-peer manner to learn the information extracted from their local data. As local PCAs are trained with only normal data (i.e., without faults), these models become sensitive to data that indicate anomalies, enabling the detection of faults. Moreover, a scrambling technique is applied to shuffle the order of the dataset, hiding the structural dependency among samples from malicious agents. Combining decentralized FL with the scrambling technique can enhance data confidentiality and cope with network scalability, as the processing of the dataset will be distributed over several nodes, hindering the access of malicious agents. Results on a testbed-derived dataset show no penalties for adopting the proposed disaggregated solution, i.e., the performance is the same as that of the centralized solutions.
