With the AI boom, data center interconnects require higher symbol rates and throughput. Intensity modulation and direct detection (IM/DD) is a cost-effective candidate for short-reach optical links. Ring resonator-based modulators (RRMs) with broad bandwidth have emerged as a promising solution for high-speed IM/DD links. However, RRMs introduce nonlinear impairments, which can cause rate bottlenecks in IM/DD. In this work, we demonstrate up to 256 GBaud RRM-based optical amplification-free on-off keying links enabled by two NN-based equalizers. To the best of our knowledge, this work is the first RRM-based IM/DD demonstration of a 256 GBaud OOK link. We achieve bit-errorratio below the 7% overhead hard-decision forward error correction threshold after transmission over 100 m single-mode fiber (SMF) for all considered symbol rates. Our work outperforms recent state-of-the-art RRM-based IM/DD studies by applying NN equalization using a 42-GHz bandwidth RRM with a driving voltage of 2.7-V<inf>pp</inf>. The achieved performance and complexity analysis reported in this work are a key step towards future implementations of NN-based equalizers in hardware to enable high-symbol-rate IM/DD systems.

Error vector magnitude (EVM) is commonly used for evaluating the quality of m-ary quadrature amplitude modulation (mQAM) signals. Recently proposed deep learning techniques for EVM estimation extend the functionality of conventional optical performance monitoring (OPM). In this article, we evaluate the tolerance of our developed EVM estimation scheme against various impairments in coherent optical systems. In particular, we analyze the signal quality monitoring capabilities in the presence of residual in-phase/quadrature (IQ) imbalance, fiber nonlinearity, and laser phase noise. We use feedforward neural networks (FFNNs) to extract the EVM information from amplitude histograms of 100 symbols per IQ cluster signal sequence captured before carrier phase recovery. We perform simulations of the considered impairments, along with an experimental investigation of the impact of laser phase noise. To investigate the tolerance of the EVM estimation scheme to each impairment type, we compare the accuracy for three training methods: 1) training without impairment, 2) training one model for all impairments, and 3) training an independent model for each impairment. Results indicate a good generalization of the proposed EVM estimation scheme, thus providing a valuable reference for developing next-generation intelligent OPM systems.

Error vector magnitude (EVM) is a metric for assessing the quality of m-ary quadrature amplitude modulation (mQAM) signals. Recently proposed deep learning techniques, e.g., feedforward neural networks (FFNNs) -based EVM estimation scheme leverage fast signal quality monitoring in coherent optical communication systems. Such a scheme estimates EVM from amplitude histograms (AHs) of short signal sequences captured before carrier phase recovery (CPR). In this work, we explore further complexity reduction by proposing a simple linear regression (LR) -based EVM monitoring method. We systematically compare the performance of the proposed method with the FFNN-based scheme and demonstrate its capability to infer EVM from an AH when the modulation format information is known in advance. We perform both simulation and experiment to show that the LR-based EVM estimation method achieves a comparable accuracy as the FFNN-based scheme. The technique can be embedded with modulation format identification modules to provide comprehensive signal information. Therefore, this work paves the way to design a fast-learning scheme with parsimony as a future intelligent OPM enabler.

Robustness against the large linewidth semiconductor laser-induced impairments in coherent systems is experimentally demonstrated for a feedforward neural network-enabled EVM estimation scheme. A mean error of 0.4% is achieved for 28 Gbaud square and circular QAM signals and linewidths up to 12.3 MHz. 

Connection provisioning in Wavelength Division Multiplexing (WDM) networks needs to account for a number of crucial parameters. On the one hand, operators need to ensure the connection availability requirements defined in Service Level Agreements (SLAs). This is addressed by selecting an appropriate amount of backup resources and recovery strategies for the connections over which services are provisioned. Services requiring less strict availability requirements can be routed over unprotected lightpaths. Services with more strict availability requirements are provisioned over protected lightpaths in order to cope with possible failures in the network. Another important aspect to consider during the provisioning process is energy efficiency. Green strategies leverage on setting network devices in Sleep Mode (SM) or Active Mode (AM) depending on whether or not they are needed to accommodate traffic. However, frequent power state changes introduce thermal fatigue which in turn has a negative effect on the device lifetime. Finally, in multi-period traffic scenarios, it is also important to minimize the number of reconfigurations of lightpaths already established in the network in order to avoid possible traffic disruptions at higher layers. The work presented in this paper tackles the connection provisioning paradigm in an optical backbone network with a multi-period traffic scenario. More specifically the paper looks into the interplay among (i) energy efficiency, (ii) thermal fatigue, and (iii) lightpath reconfiguration aspects. To this end, the Energy and Fatigue Aware Heuristic with Unnecessary Reconfiguration Avoidance (EFAH-URA) is introduced, showing that it is possible to balance the three aspects mentioned above in an efficient way. When compared to the pure energy-aware strategies, EFAH-URA significantly improves the average connection availability for both unprotected and protected connections. On the other hand, it is done at the expense of reduced energy saving.

The paper addresses the detection of malicious attacks targeting service disruption at the optical layer as a key prerequisite for fast and effective attack response and network recovery. We experimentally demonstrate the effects of signal insertion attacks with varying intensity in a real-life scenario. By applying data analytics tools, we analyze the properties of the obtained dataset to determine how the relationships among different optical performance monitoring (OPM) parameters of the signal change in the presence of an attack as opposed to the normal operating conditions. In addition, we evaluate the performance of an unsupervised learning technique, i.e., a clustering algorithm for anomaly detection, which can detect attacks as anomalies without prior knowledge of the attacks. We demonstrate the potential and the challenges of unsupervised learning for attack detection, propose guidelines for attack signature identification needed for the detection of the considered attack methods, and discuss remaining challenges related to optical network security.

Content Delivery Networks (CDNs) are crucial for enabling delivery of services that require high capacity and low latency, primarily through geographically-diverse content replication. Optical networks are the only available future-proof technology that meets the reach and capacity requirements of CDNs. However, the underlying physical network infrastructure is vulnerable to various security threats, and the increasing importance of CDNs in supporting vital services intensifies the concerns related to their robustness. Malicious attackers can target critical network elements, thus severely degrading network connectivity and causing large-scale service disruptions. One way in which network operators and cloud computing providers can increase the robustness against malicious attacks is by changing the topological properties of the network through infrastructure upgrades. This work proposes a framework for CDN infrastructure upgrade that performs sparse link and replica addition with the objective of maximizing the content accessibility under targeted link cut attacks. The framework is based on a newly defined content accessibility metric denoted as mu-ACA which allows the network operator to gauge the CDN robustness over a range of attacks with varying intensity. Two heuristics, namely Content-Accessibility Aware Link Addition Heuristic (CAA-LAH), and Content-Accessibility-Aware Replica Addition Heuristic (CAA-RAH) are developed to perform strategic link and replica placement, respectively, and hamper attackers from disconnecting users from the content even in severe attack scenarios. Extensive experiments on real-world reference network topologies show that the proposed framework effectively increases the CDN robustness by adding a few links or replicas to the network.

Network slicing enables an infrastructure provider (InP) to support heterogeneous 5G services over a common platform (i.e., by creating a customized slice for each service). Once in operation, slices can be dynamically scaled up/down to match the variation of their service requirements. An InP generates revenue by accepting a slice request. If a slice cannot be scaled up when required, an InP has to also pay a penalty (proportional to the level of service degradation). It becomes then crucial for an InP to decide which slice requests should be accepted/rejected in order to increase its net profit. This paper presents a slice admission strategy based on reinforcement learning (RL) in the presence of services with different priorities. The use case considered is a 5G flexible radio access network (RAN), where slices of different mobile service providers are virtualized over the same RAN infrastructure. The proposed policy learns which are the services with the potential to bring high profit (i.e., high revenue with low degradation penalty), and hence should be accepted. The performance of the RL-based admission policy is compared against two deterministic heuristics. Results show that in the considered scenario, the proposed strategy outperforms the benchmark heuristics by at least 23%. Moreover, this paper shows how the policy is able to adapt to different conditions in terms of 1) slice degradation penalty versus slice revenue factors, and 2) proportion of high versus low priority services.

Failure prediction based on the anomaly detection/forecasting is becoming a reality thanks to the introduction of machine learning techniques. The orchestration layer can leverage on this new feature to proactively reconfigure cloud services that might find themselves traversing an element that is about to fail. As a result, the number of cloud service interruptions can be reduced with beneficial effects in terms of cloud service availability. Based on the above intuition, this paper presents an orchestration strategy for optical cloud networks able to reconfigure vulnerable cloud services (i.e., the ones that would be disrupted if a predicted failure happens) before an actual failure takes place. Simulation results demonstrate that, with a single link failure scenario, proactive restoration can lead to up to 97% less cloud services having to be relocated. This result brings considerable benefits in terms of cloud service availability, especially in low load conditions. It is also shown that these improvements come with almost no increase in the cloud service blocking probability performance,i.e., resource efficiency is not impacted.