Search

Artificial intelligence (AI) and machine learning (ML) continue to demonstrate substantial capabilities in solving a wide range of optical-network-related tasks such as fault management, resource allocation, and lightpath quality of transmission (QoT) estimation. However, the focus of the research community has been centered on ML models' predictive capabilities, neglecting aspects related to models' understanding, i.e., to interpret how the model reasons and makes its predictions. This lack of transparency hinders the understanding of a model's behavior and prevents operators from judging, and hence trusting, the model's decisions. To mitigate the lack of transparency and trust in ML, explainable AI (XAI) frameworks can be leveraged to explain how a model correlates input features to its outputs. In this paper, we focus on the application of XAI to lightpath QoT estimation. In particular, we exploit Shapley additive explanations (SHAP) as the XAI framework. Before presenting our analysis, we provide a brief overview of XAI and SHAP, then discuss the benefits of the application of XAI in networking and survey studies that apply XAI to networking tasks. Then, we model the lightpath QoT estimation problem as a supervised binary classification task to predict whether the value of the bit error rate associated with a lightpath is below or above a reference acceptability threshold and train an ML extreme gradient boosting model as the classifier. Finally, we demonstrate how to apply SHAP to extract insights about the model and to inspect misclassifications. [ABSTRACT FROM AUTHOR]

The recent acceleration in fiber-to-the-home deployment worldwide along with the emerging 5G communications are pressuring network operators to enhance their networks to serve these new deployments. Hence, operators are seeking new high-capacity optical-network architectures, while averting excessive capital and operational expenditures. Filterless optical networks (FONs), by replacing costly wavelength selective switches in switching nodes with passive optical power splitters/combiners, currently represent a prominent candidate for cost-effective optical-network deployment. In this tutorial, we provide an overview of the architecture and the design issues of FONs when deployed in core and in metro networks. We also perform a techno-economic study to quantify the economic benefits of FONs, comparing their cost to that of state-of-the-art filtered optical networks, and we discuss how several networking problems such as resource allocation, network slicing, and protection are tackled in the context of FONs. Finally, we present our vision of how research on FONs will evolve in the coming years. [ABSTRACT FROM AUTHOR]

Optical network failure management (ONFM) is a promising application of machine learning (ML) to optical networking. Typical ML-based ONFM approaches exploit historical monitored data, retrieved in a specific domain (e.g., a link or a network), to train supervised ML models and learn failure characteristics (a signature) that will be helpful upon future failure occurrence in that domain. Unfortunately, in operational networks, data availability often constitutes a practical limitation to the deployment of ML-based ONFM solutions, due to scarce availability of labeled data comprehensively modeling all possible failure types. One could purposely inject failures to collect training data, but this is time consuming and not desirable by operators. A possible solution is transfer learning (TL), i.e., training ML models on a source domain (SD), e.g., a laboratory testbed, and then deploying trained models on a target domain (TD), e.g., an operator network, possibly fine-tuning the learned models by re-training with few TD data. Moreover, in those cases when TL re-training is not successful (e.g., due to the intrinsic difference in SD and TD), another solution is domain adaptation, which consists of combining unlabeled SD and TD data before model training. We investigate domain adaptation and TL for failure detection and failure-cause identification across different lightpaths leveraging real optical SNR data. We find that for the considered scenarios, up to 20% points of accuracy increase can be obtained with domain adaptation for failure detection, while for failure-cause identification, only combining domain adaptation with model re-training provides significant benefit, reaching 4%–5% points of accuracy increase in the considered cases. [ABSTRACT FROM AUTHOR]

Network connectivity, i.e., the reachability of any network node from all other nodes, is often considered as the default network survivability metric against failures. However, in the case of a large-scale disaster disconnecting multiple network components, network connectivity may not be achievable. On the other hand, with the shifting service paradigm towards the cloud in today's networks, most services can still be provided as long as at least a content replica is available in all disconnected network partitions. As a result, the concept of content connectivity has been introduced as a new network survivability metric under a large-scale disaster. Content connectivity is defined as the reachability of content from every node in a network under a specific failure scenario. In this work, we investigate how to ensure content connectivity in optical metro networks. We derive necessary and sufficient conditions and develop what we believe to be a novel mathematical formulation to map a virtual network over a physical network such that content connectivity for the virtual network is ensured against multiple link failures in the physical network. In our numerical results, obtained under various network settings, we compare the performance of mapping with content connectivity and network connectivity and show that mapping with content connectivity can guarantee higher survivability, lower network bandwidth utilization, and significant improvement of service availability. [ABSTRACT FROM AUTHOR]

Traffic in optical backbone networks is evolving rapidly in terms of type, volume, and dynamicity following the rapid growth of cloud-based services, ongoing adoption of 5G communications, and explosion of the Internet of Things (IoT). The elastic optical network (EON), by adopting a flexible grid, can provide the required capacity and flexibility to handle these rapid changes. However, operators rarely perform greenfield deployments, so to limit upfront investment, a gradual migration from fixed-grid to flexible-grid switching equipment is preferable. For gradual migration, switching nodes can be upgraded (starting from bottleneck network links) while keeping the rest of the traditional fixed-grid network operational. We refer to the coexistence of fixed-grid and flex-grid optical equipment as a "mixed-grid" network. Traditional algorithms for dynamic resource assignment in EONs will not effectively be applicable in a mixed-grid network due to interoperability issues among fixed- and flex-grid nodes. In this study, we propose a new algorithm, called Mixed-grid-aware Dynamic Resource Allocation, to solve the route, spectrum, and modulation-format allocation problem in a mixed-grid network while considering interoperability constraints. Our numerical results (on representative network topologies) show that the proposed method achieves 50% less blocking (for 50% offered load) compared to the traditional approach. [ABSTRACT FROM AUTHOR]

Centralized radio access networks (C-RANs) have recently been proposed to cope with the unprecedented requirements of future 5G services, in terms of network capacity, latency, service availability, and network coordination. C-RANs are based on the idea of separating baseband signal processing from radio units (RUs), namely, antenna sites in the mobile network, in such a way that baseband processing can eventually be concentrated in common locations, the central units (CUs), which can be shared among several RUs. Although C-RAN brings significant CapEx/OpEx savings, it also requires transport of high-capacity and low-latency front-haul traffic. Hence, due to the highly dynamic nature of mobile traffic, proper placement of CUs in the optical accessaggregation network should adapt to spatio-temporal traffic variation while maintaining a high degree of RAN centralization and low service blocking. In this paper, we provide an adaptive latency-aware algorithm for dynamic CU placement in optical access-aggregation networks, which targets the minimization of the number of CUs and also preforms grooming, routing, and wavelength assignment (GRWA) for mobile network traffic demands. When given the possibility to perform CU handover, i.e., to move CUs even when they are active, our algorithm, also in high load situations, provides a low number of CUs compared with fixed CU placement and keeps the blocking probability within an acceptable range. [ABSTRACT FROM AUTHOR]

Cloud radio access networks (CRANs) make it possible to reduce power consumption in future 5G networks by decoupling baseband units (BBUs) from cell sites and centralizing the baseband processing from remote radio-heads (RRHs) in BBU pools in a cloud. Although this centralization can enable power savings, it imposes much higher traffic on the optical transport network used to connect RRHs to the BBU pool, i.e., the fronthaul. In this paper, we propose a hybrid cloud-fog RAN (CF-RAN) architecture that resorts to fog computing and to network function virtualization to replicate the processing capacity of a CRAN in local fog nodes closer to the RRHs that can be activated on demand to process surplus fronthaul/cloud traffic. We devise an integer linear programming (ILP) formulation and graph-based heuristics to decide when to activate fog nodes and how to dimension wavelengths on a time-and-wavelength division multiplexing passive optical network to support the fronthaul. Our results show that our architecture can consume up to 96% less energy than a traditional distributed RAN, providing a maximum transmission latency of about 20 μs between RRHs and BBUs even in large traffic scenarios. Moreover, we demonstrate that our graph-based heuristics can achieve the same optimal solutions of the ILP formulation but with a reduction of 99.86% in the execution time. [ABSTRACT FROM AUTHOR]

To cope with the explosion of mobile traffic demand, cloud radio access networks (C-RAN) have been recently proposed. In C-RANs, the two main elements of traditional base stations (BSs), i.e., the baseband unit (BBU) and the remote radio head (RRH), are separated, and the BBU can be virtualized and located in a remote location together with BBUs of other BSs. High-bandwidth digitized baseband signals must be exchanged between BBUs and RRHs through a large-capacity fronthaul network. Wavelength division multiplexing (WDM) technology is an efficient solution to provide sufficient fronthaul bandwidth with a large number of optical transceivers. With the fixed connections between BBUs and RRHs in the traditional C-RAN, the utilization of optical transceivers as well as BBUs is inefficient. In this paper, we propose a decomposed arrayed waveguide grating router (AWGR)-enabled passive WDM fronthaul, and we formulate the minimal energy consumption problem as integer linear programming, with the objective of minimizing the energy consumption caused by optical transceivers and BBUs, etc. We also design two dynamic lightpath adjustment heuristics by light-path reconfiguration to improve energy efficiency. In addition, we take AWGR decomposition into consideration to reduce the tuning range of tunable transceivers and free spectrum range of AWGR, hence lowering the deployment cost of transport network components. Numerical evaluation shows that our proposed schemes achieve significant energy savings compared to fixed lightpath allocation. [ABSTRACT FROM AUTHOR]

Modular design based on spatial division multiplexing switches is a promising way to improve the capacity and reduce the cabling complexity of data center networks. However, due to the coexistence of mice and elephant flows, in modular data center networks a trade-off between the blocking probability and total throughput arises. In fact, blocking elephant flows would lead to a relevant penalty on the throughput but a small penalty on the blocking probability. In this paper, we investigate the relation between the blocking and throughput in modular data center networks based on optical spatial division multiplexing. We combine the two metrics linearly by a weight factor that prioritizes them relatively. To solve the resource allocation problem, we propose both mixed integer linear programming formulations and close-to-optimal heuristics for three different spatial division multiplexing switching schemes. Simulation results demonstrate that a carefully chosen weight factor is necessary to achieve a proper balance between the blocking probability and throughput for all the schemes. [ABSTRACT FROM AUTHOR]

Mobile fronthaul (MF) provides connectivity between a remote radio head (RRH) and base-band processing unit (BBU) in a cloud radio access network. Dedicated MF connections between RRH and BBU are costly due to high bandwidth requirements. The cost issue becomes more pressing as cell densification increases the number of connections. A time-division multiplexing Ethernet passive optical network (TDM-EPON) is a promising solution to reduce MF cost, as it can multiplex several fronthaul flows in the MF network. However, even though mobile users transmit intermittently, RRH samples radio signals continuously, resulting in large constant line-rate connections between RRH and BBU that require huge aggregated bandwidth and render statistical multiplexing gain marginal. To improve upstream transmission efficiency by utilizing statistical multiplexing in an MF network, we propose to enhance the conventional TDM-EPON architecture. The enhanced architecture employs two techniques: (1) traffic classification (classifying upstream traffic into useful and useless by using machine-learning classifiers); and (2) useless-data sifting (sifting useless traffic to avoid the transmission of unnecessary EPON frames). We explore two feature-selection methods together with classifier biasedness in the first technique to achieve the best classification results, which then serve as input to our proposed sifting-based hybrid bandwidth allocation scheme in the second technique. Simulation results show significant improvements in terms of per-RRH traffic load, number of RRHs supported by same EPON, and signal-to-noise ratio, while keeping the end-to-end delay under 100 μs. [ABSTRACT FROM AUTHOR]

A scalable, low-latency, high-speed, and energy- efficient data center network is a key element in the deployment of future large-scale data centers, and photonic switching has recently been recognized as a promising solution to fulfill these goals. In this study, we present a packet-switched optical network (PSON) architecture with centralized control for intra-data-center connectivity. For efficient PSON operation, intelligent yet low-complexity bandwidth-scheduling algorithms are critical. To align with realistic traffic flows in a data center, we consider ?mice flow,? which occurs frequently but carries a small number of bytes, and ?elephant flow,? which occurs occasionally but has a huge number of bytes. To classify traffic flows with different characteristics, we investigate various machine-learning (classification) techniques, such as C4.5 and Na?ve Bayes Discretization, and compare their performance in terms of accuracy and classification speed. We also develop a priority-aware scheduling algorithm for packet switching, which is optimized for PSON, and is adaptive to flow classification under a dynamic traffic scenario. Numerical simulations show that our proposed scheduling algorithm assisted by flow-classification techniques can outperform a benchmark algorithm in terms of average delay and packet-loss ratio. [ABSTRACT FROM PUBLISHER]

Bandwidth requirements of both wireless and wired clients in access networks continue to increase rapidly, primarily due to the growth of video traffic. Application awareness can be utilized in access networks to optimize quality of experience (QoE) of end clients. In this study, we utilize information at the client-side application (e.g., video resolution) to achieve superior resource allocation that improves user QoE. We emphasize optimizing QoE of the system rather than quality of service (QoS), as user satisfaction directly relies on QoE and optimizing QoS does not necessarily optimize QoE, as shown in this study. We propose application-aware resource-allocation schemes on an Ethernet passive optical network (EPON), which supports wireless (utilizing orthogonal frequency division multiple access) and wired clients running video-conference applications. Numerical results show that the application-aware resource-allocation schemes improve QoE for video-conference applications for wired and wireless clients. [ABSTRACT FROM PUBLISHER]

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. [ABSTRACT FROM PUBLISHER]

Although a promising solution for next-generation radio access networks (RANs), cloud RAN (C-RAN) still faces technical challenges due to the separate locations of the cloud-based baseband unit (BBU) pool and remote radio heads (RRHs). Fortunately, time- and wavelength-division multiplexing passive optical networks (TWDM-PONs) can efficiently bridge the communications between the BBU pool and RRHs. In this work, we address the problem of formulating virtual time division multiplexing PONs (vTDM-PONs) adaptively in a TWDM-PON-based C-RAN. However, instead of relying on a centralized approach, we develop algorithms that can work in a distributed manner. Specifically, we consider a scenario in which the optical network unit (ONU) of each RRH has the initiative to choose a proper vTDM-PON to register to based on its knowledge on the C-RAN. We formulate game theoretic models for the vTDM-PON formation for both static and dynamic operation of a C-RAN, then transform the games into weighted potential games and therefore prove the existence of Nash equilibrium (NE) points in the games. Moreover, we propose a distributed algorithm that can converge to the NE point(s) to enable each ONU to choose its vTDM-PON intelligently. As a comparison, we formulate a centralized mixed-integer linear programming model to provide the optimum vTDM-PON formation solution. Simulation results confirm the effectiveness of the design of the proposed game models and distributed algorithm. [ABSTRACT FROM PUBLISHER]

Recently, the IEEE 802.3 Ethernet Working Group has classified three architectures for the nextgeneration Ethernet passive optical network (NG-EPON). They are called single-scheduling domain (SSD) PON, multi-scheduling domain (MSD) PON, and wavelength-agile (WA) PON, and they differ based on how a group of optical network units (ONUs) share awavelength. Existing dynamic bandwidth and wavelength allocation (DBWA) schemes for conventional EPON can be applied to MSD-PON and SSDPON, but not WA-PON. This is because WA-PON is a new architecture with full flexibility where a flexible number of wavelengths can be assigned to one ONU, and multiple ONUs can transmit at the same time. In this work, we develop a mathematical model and a novel DBWA scheme for transmission scheduling in WA-PON. However, as WA-PON incurs penalties in terms of delay and power consumption when an ONU activates its transmissions on new wavelengths, a trade-off between energy saving and data-transfer latency reduction needs to be carefully addressed when performing transmission scheduling. So, we develop a power-consumption model and modify the proposed DBWA scheme to enhance the energy efficiency ofWA-PON. Finally, we conduct simulation experiments for performance evaluation of the three PON architectures in terms of latency and packet loss ratio.We quantitatively investigate the influence of various parameters, such as the number of ONU transceivers and ONU buffer size, onWA-PON latency and packet loss ratio, and we evaluate the energy efficiency gain of the modified DBWA scheme. [ABSTRACT FROM PUBLISHER]

Using virtualization, service providerscan create an abstraction of the physical servers hosted in their data centers, and run their services directly on these abstract entities, called virtual machines (VMs). As a direct result of service virtualization, services hosted on VMs can be migrated from one data center to another. VMmigration can be performed without or with a minimal service interruption, and in this case we talk about live VM migration. The main issue of live VM migration is that memory pages are modified during the process. Due to this fact, the same memory pages might be transferred several times, thus increasing both the migration duration and the amount of network resources occupied by the VM migration process. Considering the increasing amount of VM migrations carried on in inter-data-center networks, efficient strategies for live-VM-migration bandwidth provisioning are needed. In this paper, we propose and compare various novel routing and bandwidth assignment algorithms for the live migration of VMs in a distributed DC infrastructure, under dynamic traffic conditions. We find that assigning a bandwidth to theVMmigration request in proportion to the path length with the objective of minimizing the amount of resources occupied by the migration request could improve the network performance. In addition,wepresent a comparative analysis of the serial and the parallel migration of multiple VMs.We quantify the impact of the migration strategy (serial versus parallel) on blocking probability, migration duration, downtime, and resource occupation. [ABSTRACT FROM PUBLISHER]

Common Public Radio Interface (CPRI) is a successful industry cooperation defining the publicly available specification for the key internal interface of radio base stations between the radio equipment control (REC) and the radio equipment (RE) in the fronthaul of mobile networks. However, CPRI is expensive to deploy, consumes large bandwidth, and currently is statically configured. On the other hand, an Ethernet-based mobile fronthaul will be cost-efficient and more easily reconfigurable. Encapsulating CPRI over Ethernet (CoE) is an attractive solution, but stringent CPRI requirements such as delay and jitter are major challenges that need to be met to make CoE a reality. This study investigates whether CoE can meet delay and jitter requirements by performing FPGA-based Verilog experiments and simulations. Verilog experiments show that CoE encapsulation with fixed Ethernet frame size requires about tens of microseconds. Numerical experiments show that the proposed scheduling policy of CoE flows on Ethernet can reduce jitter when redundant Ethernet capacity is provided. The reduction in jitter can be as large as 1 μs, hence making Ethernet-based mobile fronthaul a credible technology. [ABSTRACT FROM PUBLISHER]

Modular design has recently emerged as an efficient solution to build large data center (DC) facilities. Modular DCs are based on stand-alone prefabricated modules (i.e., PODs) that can be easily installed and interconnected. PODs can generate a large amount of traffic and thus require an ultra-high-capacity interconnection network. However, current electronic and optical interconnect architectures applied to modular DCs may experience major scalability problems in terms of high energy consumption and cabling complexity. To address these problems, we investigate five optical interconnect architectures based on spatial division multiplexing (SDM), and for each architecture, we propose a resource allocation strategy. We also present an extensive comparison among the SDM architectures in terms of cost and performance (i.e., blocking probability and throughput), with the objective to find the architecture offering the best trade-off between cost and performance for given DC sizes and traffic load values. Our results demonstrate that, in small modular DCs with low traffic load, an architecture based only on SDM is the best option, while in medium DCs with medium traffic load, an architecture based on coupled SDM and flexgrid wavelength division multiplexing (WDM) with spectral flexibility is the best solution. Finally, for large DCs with high traffic load values, the best trade-off between cost and performance is achieved by an SDM architecture that is based on uncoupled SDM and flexgrid WDM. [ABSTRACT FROM PUBLISHER]

To meet challenging 5G capacity requirements, operators are densifying their cellular networks by deploying additional small cells to cover hot spots, and such an increase in the number and density of cells may result in excessive numbers of handovers. In this study, we propose a handover reduction mechanism implemented in a cloud radio access network (CRAN) by exploiting the high capacity of an optical access network serving as a "fronthaul." CRAN has been proposed as a 5G radio access network architecture, where the digital unit (DU) of a conventional base station (BS) is separated from the radio unit (RU) and moved to the "cloud" (DU-cloud) for better mobility management and cost saving. Separating RUs and DUs requires a low-latency and high-bandwidth 5G transport network to handle "fronthaul" traffic, for which optical access is an excellent choice. Here, we present a new 5G architecture, called virtualized-CRAN (V-CRAN), moving toward a cell-less 5G mobile network architecture. We leverage the concept of a "virtualized-BS" (V-BS) that can be formed by exploiting several enabling technologies such as software-defined radio and coordinated multipoint transmission/reception. A V-BS can be formed on a per-user basis by allocating virtualized resources on demand so that common signals can be jointly transmitted from multiple RUs to the user without triggering handover. We first model the handover reduction optimization problem for a scenario where future mobility information is known, and then propose a suite of algorithms for a scenario where future information is unknown. Simulation results show that V-CRAN can enhance the throughput of users at the cell-edge, as well as significantly reduce the number of handovers, handover delay, and failure rate. [ABSTRACT FROM PUBLISHER]

Resource crunch, i.e., severe reduction of network resources caused by large-scale failures or drastic traffic fluctuations, represents a crucial concern for network survivability. Under resource crunch, simply relying on preserving large amounts of redundant resources to achieve network robustness becomes difficult or even impossible. However, reactively optimizing network resources and re-provisioning disrupted and/or existing connections might be a promising solution for disaster failures due to earthquake, tsunami, hurricane, malicious attack, etc. In this paper, we study a re-provisioning method for survivable networks, namely, a fairness-aware degradation based multipath re-provisioning (FDM) strategy to combat large-scale cascading and/or correlated failures in post-disaster telecom mesh networks (such as optical backbone networks). By exploiting policies of fairness-aware bandwidth degradation and multipath deployment, FDM exploits the potential bandwidth resource in post-disaster networks to maintain network connectivity and maximize traffic flows, while avoiding temporary interruptions of existing connections and balancing the traffic distribution. A mixed integer linear programming model and a heuristic algorithm, called FDM-i and FDM-h, respectively, are developed and applied in different scale test networks under various volumes of post-disaster traffic demands. Simulation results show that, with respect to some counterparts, the proposed FDM schemes achieve better performance in terms of connection loss ratio, traffic loss ratio, and fairness of traffic distribution and demonstrate a good trade-off between resource utilization and traffic distribution. Compared with FDM-i, FDM-h achieves similar performance and shows better characteristics on adaptability and scalability. [ABSTRACT FROM PUBLISHER]

In this paper, we focus on the survivability of elastic optical networks (EONs) that jointly support two types of traffic demands: unicast and anycast. To provide network survivability, we apply multipath routing; i.e., we allow the splitting of a demand into a number of routing paths if the paths' combination guarantees the realization of a specific demand volume in the case of a single link failure. We formulate the corresponding optimization problem as an integer linear program (ILP) and propose a survivable multipath allocation (SMA) algorithm to solve the problem in a reasonable amount of time. Next, we perform numerical experiments to compare the efficiency (ability to provide a good-quality solution in a reasonable amount of time) of the ILP model and SMA as well as to evaluate the impact of survivable multipath routing on the objective defined as a maximum spectrum usage in EONs. Our results show that the SMA method finds good-quality solutions in a reasonable amount of time and that survivable multipath routing in EONs requires additional spectrum resources, up to 45%. However, the amount of additional resources depends on the required protection level, amount of anycast traffic, the maximum number of paths used for demand realization, and the considered network topology. [ABSTRACT FROM PUBLISHER]

The growing popularity of high-speed mobile communications, cloud computing, and the Internet of Things (IoT) has reinforced the tidal traffic phenomenon, which induces spatio-temporal disequilibrium in the network traffic load. The main reason for tidal traffic is the large-scale population migration between business areas during the day and residential areas during the night. Traffic grooming provides an effective solution to aggregate multiple fine-grained IP traffic flows into the optical transport layer by flexibly provisioning lightpaths over the physical topology. In this paper, we introduce a comprehensive study on energy efficiency and network performance enhancement in the presence of tidal traffic. We propose and leverage the concept of stateful grooming to apply differentiated provisioning policies based on the state of network nodes. We formulate and solve the node-state-decision optimization problem, which can decide the specific state of network nodes when a certain traffic profile is given, considering the trade-off between energy efficiency and blocking performance with a multi-objective integer linear program (MOILP). Then, we propose an online traffic-aware intelligent differentiated allocation of lightpaths (TIDAL) algorithm, based on stateful grooming and the MOILP, to accommodate the dynamic tidal traffic. Our illustrative numerical results show that the proposed method can achieve a significant performance improvement in an energy-efficient way. [ABSTRACT FROM PUBLISHER]

Cloud services based on content sharing and distribution are most prevalent in today's world, and datacenter (DC) networks supporting such services manage a huge volume of data transfers, in the range of terabytes to zettabytes. Usually, contents are replicated in geographically distributed DCs for high availability and reliability, but in the case of high-impact, large-scale disasters, like weapons of mass destruction attacks, DC networks can suffer massive service disruptions and data loss. To save critical data under such circumstances, contents could be evacuated in response to an upcoming disaster alert from a likely disaster region to a safe location before the disaster occurs and causes serious data damage. Depending on the forecasted disaster scenario, content evacuation can be greatly constrained by limited available network resources and strict deadlines (evacuation times). We propose a rapid-data-evacuation (RDE) heuristic that selects the least-delay paths (considering propagation delays, network bandwidth, and congestion) through an anycast network model, and schedules critical and vulnerable contents for evacuation such that the maximum amount of contents can be evacuated within the evacuation deadline or equivalently, a given amount of contents can be evacuated in minimum time. We compare our RDE heuristic with a nearest-evacuation approach, which evacuates data only to the nearest DC using the shortest path. Our results show that, for typical scenarios considered in this study, compared to nearest evacuation, our rapid-evacuation approach provides about 64% time savings, or equivalently, about 97% more volume of evacuated contents for a given deadline. Since our algorithm is based on a greedy approach, we also present an enhanced RDE algorithm based on the simulated annealing (SA) metaheuristic as a benchmark. We show that our proposed RDE heuristic performs very close to the SA based algorithm and is much faster in computation time. Hence, it is suitable and efficient for rapid evacuation. [ABSTRACT FROM PUBLISHER]

Recent targeted attacks and natural disasters have made disaster-resilient cloud network design an important issue. Network operators are investigating proactive and reactive measures to prevent huge data loss and service disruptions in case of a disaster. We present novel techniques for disaster-aware datacenter placement and content management in cloud networks that can mitigate such loss by avoiding placement in given disaster-vulnerable locations. We first solve a static disaster-aware datacenter and content placement problem by adopting an integer linear program with the objective to minimize risk, defined as expected loss of content. It is a measure of how much, in terms of cost or penalty, a network operator may lose probabilistically due to possible disasters in a cloud network. We also show how a service provider's budget constraint can affect disaster-aware placement design. Since disaster scenarios, content popularity, and/or importance are always changing in time, content placement should rapidly adapt to these changes. We propose a disaster-aware dynamic content-management algorithm that can adjust the existing placement based on dynamic settings. Besides reducing the overall risk and making the network disaster-aware, reducing network resource usage and satisfying quality-of-service requirements can also be achieved in this approach. We also provide a cost analysis of employing a dynamic disaster-aware placement design in the network based on real-world cloud pricing. [ABSTRACT FROM PUBLISHER]

Telecommunication networks face large-scale threats, such as natural disasters (earthquake, hurricane, tsunami, etc.) and targeted attacks (weapons of mass destruction). These disasters may simultaneously destroy many network components (nodes and/or links) and cause multi-failure events. Although existing protection schemes can handle some given multi-failure scenarios, it is extremely challenging to solve all possible disaster-failure scenarios with reasonable resource costs and a 100% reliability guarantee. In this paper, we investigate postdisaster restoration strategies for telecom backbone mesh networks (such as wavelength-division multiplexed networks), whose objective is two-fold: to maintain network connectivity (i.e., guarantee a minimal degree of service for the survived connections) and to maximize the traffic flow in the post-disaster network. We study three postdisaster re-provisioning schemes employing different policies for connection rerouting and admissible bandwidth degradation (i.e., provisioning partial bandwidth). These schemes, whose services range from essential to global post-disaster re-provisioning, are the no-degradation reprovisioning (NDR) scheme, the degradation-as-needed re-provisioning (DAN) scheme, and the fairness-aware degradation re-provisioning (FAD) scheme. Corresponding mixed integer linear program models are developed and applied on two representative mesh topologies. Numerical results show that NDR can effectively decrease the connection loss ratio by rerouting some survived connections. However, both DAN and FAD can achieve optimal performance on maintaining connectivity even without rerouting any survived connections. DAN outperforms the other schemes in terms of traffic loss ratio, but it may cause a large difference between the maximum and minimum degradation ratios. FAD can improve the performance in terms of traffic loss ratio by rerouting a few survived connections. In particular, compared with DAN, FAD can provide balanced bandwidth degradation for all connections by sacrificing a certain amount of traffic flow, but avoiding extreme deterioration of the bandwidth capacity on some connections. [ABSTRACT FROM PUBLISHER]

Energy efficiency is an important target for the management of wavelength-division-multiplexing-based core optical networks. A possible power management strategy for energy-efficient optical networks consists in putting opto-electronic devices (transponders and regenerators) into a low power mode, called sleep or idle, or even turning them off, during low traffic periods. It has been shown that sleep-mode enabled transponders and regenerators yield to substantial energy savings; however, their nonnegligible wake-up time may degrade the network performance in terms of the blocking probability of the connections. Opaque networks, in which transponders are utilized at each node in the end-to-end path, are especially prone to such degradations since multiple network nodes may simultaneously exhaust their pool of transponders due to long wake-up times, increasing the blocking probability. On the other hand, these degradations are less severe in transparent networks. In this paper, we evaluate how the duration of the wake-up time affects the performance of the network in terms of blocking probability depending on its architecture (transparent, translucent, or opaque) and the dynamicity of the traffic. Additionally, we propose a novel routing algorithm to mitigate the blocking probability due to the wake-up time. The benefits of the proposed algorithm are highlighted through extensive results. [ABSTRACT FROM PUBLISHER]

The adoption of a flexible grid will benefit the network design and control plane of future optical networks by providing increased adaptability of spectral resources to heterogeneous network conditions. Unfortunately, this flexibility is gained at the cost of significant additional complexity in the network design and control. In this paper, we consider the optimization of routing and spectrum allocation in flexi-grid ring networks and explore the trade-off between network cost (in terms of spectrum and transponder utilization) and problem complexity (in terms of the number of variables/constraints and computational time). Such trade-offs are investigated under multiple assumptions in terms of traffic grooming, regeneration, and modulation/baud rate assignment capabilities and contrasted with the case of fixed grid.We show how in the presence of traffic grooming the additional complexity due to the flexible grid has a minor impact on problem complexity. Similarly, in all the considered scenarios, regeneration and modulation/baud rate assignment do not relevantly impact on problem complexity. We also consider two possible alternative integer linear programming (ILP) models: the slicebased and channel-based approaches. The former handles each slice individually, whereas the latter uses precomputed subsets of contiguous slices of different bandwidths. Both models are solved under several different network settings. Complexity comparison of the ILP models shows that the slice-based approach provides better performance than the channel-based approach and that the performance gap between the two models increases with the introduction of additional flexibility and dimensions. [ABSTRACT FROM PUBLISHER]

Backbone networks usually have some excess capacity (EC) to accommodate traffic fluctuations and to avoid early capacity exhaustion. Network operators can exploit EC in optical wavelength division multiplexed (WDM) backbone networks to support survivable traffic grooming, where connection requests are of subwavelength granularity and each provisioned request has to be protected from single-link failures. We investigate novel EC management techniques that can improve network performance, in terms of Service-Level Agreement (SLA) violations and bandwidth blockings, with no requirement of deploying additional capacity. We investigate exploiting and managing EC by the following techniques. i) Preprovisioning: When traffic is light, network resources are reserved by a preprovisioning scheme, i.e., a connection can be provisioned on reserved protected links to increase availability. We show that preprovisioning also decreases connection setup time, an important metric for delay-sensitive services. ii) Backup reprovisioning: Since high-availability protection schemes usually consume more resources, connections in our solution can be switched to a protection scheme that provides lower availability (but higher resource efficiency) by reprovisioning backup resources when traffic increases. iii) Hold-lightpath: We propose a new ?hold-lightpath? scheme to exploit EC. This scheme prevents the termination of pre-established (but unused) resources to increase availability and decrease connection setup time. We compare our techniques with traditional protection schemes for typical daily fluctuating traffic on typical backbone network topologies and find that significant improvements can be achieved in terms of decreasing SLA violations, bandwidth blocking, and connection setup time. [ABSTRACT FROM PUBLISHER]

Mixed-line-rate (MLR) networks have been proposed recently to support emerging traffic demands with diverse capacity requirements. In such networks, wavelength channels of the optical paths can have a variety of capacities (10/40/100 Gbps). It has been shown that this feature enables new paradigms for protection in optical wavelength-division-multiplexing (WDM) networks. In this paper, we propose to design transparent MLR networks with shared subconnection protection (SSP). SSP in an MLR network, compared with other more classical approaches (dedicated or shared protection), can also address the cost-versus-capacity trade-off in providing shared protection in a transparent network. The design problem is solved by a two-step approach that formulates part of the problem as an integer linear program (ILP). Our results show that SSP enables higher sharing of backup capacityin an MLR network in that the design cost is up to 37% less than a single-line-rate network with SSP and around 40% less than MLR with dedicated protection. [ABSTRACT FROM PUBLISHER]

Future passive optical networks (PONs) are expected to support much larger capacity and much wider coverage. How to jointly address these two design requirements represents one of the most challenging aspects in today's research on PONs. So far, most research efforts have been devoted to devising new architectural or technological solutionsto support such stringent requirements. In particular, the technology to be used for transmission may have diverse multiplexing techniques, types of transceivers, modulation formats, and detection techniques. However, the question of which transmission technology is the most effective considering the trade-offs in terms of complexity, offered capacity, and reach is still open. In this paper, we aim at answering this question comparing the use of time division multiplexing (TDM), wavelength division multiplexing (WDM), and hybrid TDM/WDM techniques in PONs. To achieve this goal, we first categorize the main options for transmission technologies in PONs in three families: colored dense WDM (DWDM), tunable DWDM, and colorless DWDM. Then, we propose a new optimization scheme that selects the optimal transmission technology for different operational scenarios that are defined by varying the number of users, the distance to the users, and the traffic load. The choice of the passive optical components to be used at the remote node is also part of the optimization scheme given its significant impact on the choiceof the transmission technology. As a result, we report and discuss which transmission technologies are the most suitable under different operational scenarios. [ABSTRACT FROM PUBLISHER]

For decades, optical networks have provided larger bandwidths than could be utilized, but with the increasing growth of the global Internet traffic demand, new optical transmission technologies are required to provide a much higher data rate per channel and to enable more flexibility in the allocation of traffic flow. Currently, researchers are investigating innovative transceiver architectures capable of dynamically adapting the modulation formatto the transmission link properties. These transceivers are referred to as elastic and enable flexible allocation of optical bandwidth resources. To exploit their capabilities, the conventional fixed spectrum grid has to evolve to provide a more scalable and flexible system that can provide the spectral resources requireded to serve the client demand. The benefits of elastic transceivers with distance-adaptive data rates have been evaluated in optical core networks, but their application to metro ring networks has still not been addressed. This paper proposes methods based on integer linear programs and heuristic approaches to solve the routing, modulation level, and spectrum assignment problem in optical rings with elastic transceivers and rate-adaptive modulation formats. Moreover, we discuss how to analytically compute feasible solutions that provide useful upper bounds. Results show a significant reduction in terms of transceiver utilization and spectrum occupation. [ABSTRACT FROM PUBLISHER]

Traffic in telecom networks is increasing rapidly, and it has been a challenging task to plan and upgrade network capacities for this increasing traffic, keeping the cost of resources within a targeted budget. Besides handling this drastic growth in traffic, future optical networks are also expected to be increasingly heterogeneous with respect to services supported and underlying technologies employed. To support this heterogeneous volume of traffic, mixed-line-rate (MLR) optical networks with line rates of 10, 40, and 100 Gbps have been shown to be effective. In the case of a greenfield network design, i.e., when planning capacities for a new network, MLR networks have been shown to be cost efficient as well as energy efficient in some recent studies. The concept of an MLR network is evolutionary, starting from a 10G single-line-rate network to a coexistence of multiple line rates in the same network as capacities on some links are periodically upgraded from 10G per wavelength to 40G and 100G per wavelength with traffic growth. Therefore, from a network-upgrade perspective, an important issue is to devise a cost-optimized migration strategy from 10G to 40G to 100G and beyond, given a traffic growth model. However, energy consumption in different elements in the network, especially in those elements whose energy consumption depends on the bandwidth of the traffic that they are handling, is also an important parameter to consider. Therefore, the ultimate question is: Can an MLR be a good candidate for energy-efficient and cost-efficient upgrade? In this study, we investigate the energy-efficient and cost-efficient MLR network-upgrade problem. In this context, we also study the effect of network connection disruption on energy-efficient and cost-efficient MLR network upgrade. In general, the service provider's aim would be to have as few disruptions as possible during capacity upgrade, as disruptions may induce service degradation. Our results show that the amount of disruptions has a conflicting effect on energy-efficient and cost-efficient upgrade in MLR networks, and we develop an optimized upgrade strategy so that both cost and energy are kept within a certain limit. [ABSTRACT FROM PUBLISHER]

This Special Issue contains a collection of outstanding papers covering several applications of machine learning in optical communications and networking, including: resource management in elastic optical networks, datacenters and ROADM-based networks; noise and signalquality estimation; and traffic prediction. We provide a brief overview of the current state of machine learning in optical networks, followed by a categorization of the eleven papers in this special issue. [ABSTRACT FROM AUTHOR]

This publisher's note corrects the Acknowledgment section in [J. Opt. Commun. Netw. 10, 749 (2018)]. [ABSTRACT FROM AUTHOR]

The ONDM 2016 Special Issue is a collection of outstanding optical networking papers, chosen among the ONDM 2016 most highly ranked papers. The papers in this special issue cover a broad range of topics including dynamic elastic optical networks, network virtualization, next-generation optical access networks, spatial division multiplexing networking, and multi-layer design. [ABSTRACT FROM PUBLISHER]

In this paper we propose a design procedure based on comprehensive integer linear programing (ILP) for translucent optical transport networks (OTNs), exploiting an extended version of the so-called “connectivity graph.” For the first time to the best of our knowledge, we propose a mathematical approach that covers the manifold challenges of a realistic optical layer OTN design, jointly solving the regenerator placement problem (RPP) and the routing, fiber and wavelength assignment with regenerator problem (RFWA-RP). As a first contribution, we extend the concept of the connectivity graph to a k-path (k-p) connectivity graph. Second, the formulation addresses the problem of planning the number of dense wavelength division multiplexing systems, jointly solving the RPP and placing transponders. Third, we consider regeneration devices as wavelength converters also, showing the benefits arising from that aspect disregarded in the existing impairment-aware (IA) ILP formulations based on the connectivity graph. Finally we compare over different networks the results of our design procedure with those achieved by a hybrid method that combines a simplified IA-ILP formulation with a greedy heuristic. Moreover, an analytical framework to evaluate the network cost in terms of capital expenditure and operational expenditure is also presented. We show that the k-p connectivity graph represents a cost-effective tool for network design, as it allows one to greatly reduce the number of resources needed in both the protected and the unprotected scenarios. [ABSTRACT FROM PUBLISHER]