The increasing complexity and stringent demands of modern communications require advanced automation since traditional methods cannot effectively address the dynamic nature of network operations. This demonstration showcases a Machine Learning Operations (MLOps) framework tailored for optical networks. It highlights the deployment and operation of deep reinforcement learning (DRL) models to tackle complex, dynamic challenges in optical networks by presenting practical use cases. Through an experimental setup, it illustrates the end-to-end lifecycle management of DRL agents integrated into an SDN control plane. The demonstration spans data collection, model training, deployment, and continuous monitoring, showcasing the unique benefits and challenges of intelligent optical network operations. The demonstration provides practical insights into deploying intelligent network functions and emphasizes the critical role of MLOps in achieving zero-touch management for future optical infrastructures.

With the advent of the multimodal immersive communication system, people can interact with each other using multiple devices for sensing, communication and/or application level control either onsite or remotely. As a breakthrough concept, a distributed sensing, computing, communications, and control (DS3C) fabric is introduced in this paper for provisioning 6G services in multi-tenant environments in a unified manner. The DS3C fabric can be further enhanced by natively incorporating intelligent algorithms for network automation and managing networking, computing, and sensing resources efficiently to serve vertical use cases with extreme and/or conflicting requirements. As such, the paper proposes a novel end-to-end 6G system architecture with enhanced intelligence spanning across different network, computing, and business domains, identifies vertical use cases and presents an overview of the relevant standardisation and pre-standardisation landscape.

We present P5; an event-driven policy framework that allows network operators to realize end-to-end policies on top of P4-based data planes in an intuitive and effective manner. We demonstrate how P5 adheres to a service-level agreement (SLA) by applying P4-based traffic engineering with latency constraints.

Orchestration of computing and network resources across multiple technology layers is critical for the autonomous roll-out of network services that fulfill the heterogeneous requirements of vertical industries. Therefore, effectively mapping the requirements of verticals into underlying infrastructure management decisions is a challenging issue. In this paper, we evaluate the provisioning of vertical-driven network services, composed of a set of interconnected virtual network functions, on top of a distributed cloud infrastructure interconnected through an Elastic Optical Network (EON). To this end, we propose a solution exploiting the advantages of Deep Reinforcement Learning. This solution relies on two cooperating agents to execute an efficient selection of computing resources in the cloud infrastructure for the placement of virtualized network functions, as well as the routing and spectrum assignment in the EON. The performance of the proposed solution is evaluated in scenarios where arrival and departure of service requests are dynamic. Experimental results show the improved performance of the proposed solution with and without invalid action masking over a heuristic algorithm in terms of service blocking rate.

The expected complexity of shared 5G/6G cloud and network infrastructures requires a functional management and orchestration (MANO) architecture to automatically provision distinct vertical services. These are mapped as generic network services (NSes) specifying their computing and networking needs: Virtual Network Functions (VNFs) and Virtual Links. We focus on a hierarchical MANO implementation where a Resource Layer orchestrator handles the configuration of the physical infrastructure and exposes an abstract view of it to the upper-layer Service Orchestrator. Two different abstraction philosophies are adopted, namely the Infrastructure Abstraction, which pre-calculates the resource allocations before advertising them, and the Connection Service Abstraction, which exposes potential connectivity services without an actual resource allocation. The resulting abstracted infrastructure becomes the input of the Service Orchestrator to make (placement) decisions fulfilling the NS demands. Thus, a novel placement algorithm is devised which, unlike previous works, can handle NSes with arbitrary VNF topology demands. The performance of the MANO functions is experimentally evaluated by dynamically creating/terminating heterogeneous NSes in terms of: NS blocking, blocked bandwidth/VNF ratio, bandwidth occupancy, and VNF distribution per cloud site. From the results, Connection Service Abstraction does attain a more efficient use of resources and in general, performs better than the Infrastructure Abstraction approach.

Slices are deployed to continuously fulfill requirements from vertical industries. Thus, the management of service level agreements (SLAs) is fundamental. Composite network services (NSs), made up of multiple nested NSs, can be exploited to build slices. Generally, slices may span multiple domains and the different nested NSs may be deployed by different providers, in what is referred to as federated NS. Previous work focused on the architectural design and workflows for the functional deployment of federated services without considering the continuous SLA management. Herein, we cover this gap by presenting a complete workflow, the required interface extensions, and the operation profiling over an experimental testbed to automatically scale nested NSs deployed in a federated domain. Evaluation results show that such scale out/in operations can be performed in tens of seconds (39/31 s, respectively), where the allocation/release of underlying resources account for most of the profiled scaling time (more than 80%).

Heterogeneity is one relevant characteristic of next generation mobile networks. This term embraces not only different kind of infrastructure resources or transmission technologies (e.g., networking vs computing, wireless vs optical), but it also applies to different ways of implementing the network functions (NFs) composing the network services (NSs). This allows the definition of hybrid NSs combining different kind of components, such as physical, virtual, and cloud-native NFs. This demonstration shows the enhancements introduced in the 5Growth platform to manage physical NFs, hence increasing the capabilities of this platform to cope with more heterogeneous hybrid NSs in multi-site scenarios. In particular, we show the deployment of an NS constituted by physical, virtual, and cloud-native NFs implementing an end-to-end service covering access, mobile core and application functionalities.

This demonstration will showcase the end-to-end orchestration of virtual network functions in the full-fledged ADRENALINE Testbed Cloud Platform expanding from the edge to the cloud. The Management and Orchestration (MANO) software ETSI OpenSource MANO (OSM) is used to deploy and handle a multi-site network service involving both edge and core Data Centers (DCs). Besides, the inter- and intra-DC connectivity is directly managed by a novel OSM WAN Infrastructure Manager (WIM) connector using the Transport API (TAPI) interface, thus completely abstracting the details of the underlying SDN controllers handling the programmability of the WAN network interconnecting the DCs.

In 5G networks, heterogeneous vertical services with different requirements are rolled out over a common multi-technology infrastructure. A resource orchestrator entity automatically coordinates the operations and functions to support the service's lifecycle management (i.e., creation, update and termination). Moreover, it is essential that service needs are continuously assured even if transport network anomalies (e.g., link failures) occur. Herein, we present an implemented resource orchestrator architecture integrating monitoring capabilities to attain closed-loop operations for: i) gathering monitored information; ii) detecting transport network anomalies; and iii) triggering the required action (e.g., restoration) to keep the service continuity. When a link failure happens, several transport services may be disrupted requiring to be immediately restored. To this end, we propose a novel on-line restoration algorithm called as Global Concurrent Optimization (GCO). The GCO algorithm aims at attaining an enhanced restorability performance compared to a more traditional restoration algorithm (referred to as 1-by-1). Both algorithms are experimentally compared on top of the deployed resource orchestrator architecture. The evaluation is done upon both dynamic service arrival/departure and link failure generation using different performance metrics: the average restorability, the average network resource utilization, and the restoration computational time.

Next generation mobile networks base on the auto- mated, flexible, and dynamic orchestration of virtualised network services (NSs). These NSs are made up of Virtual Network Functions, which have hitherto mostly been implemented by means of virtual machines. The current trend is to include the use of containers, the so-called Cloud-native Network Functions, which may fit better the need of NS deployments embracing edge infrastructures having constrained resources. This ends up in the definition of NSs mixing both kinds of network functions (NFs) satisfying the needs of network operators and vertical industries and the characteristics of available infrastructures. We refer to the use of both kinds of NFs in a single NS as Hybrid NS. This demonstration presents the extensions done in the 5Growth management and orchestration platform to perform the deployment of such kind of NSs in a multi-site infrastructure, including the dynamic interconnection of the deployed NFs according to their nature.

Measuring several relevant metrics related to Network Function Virtualization (NFV) service lifecycle management in real time brings an enhanced monitoring of the operation of the network service orchestrator (SO) and the NFV infrastructure. In this demonstration, we integrate a complete data engineering pipeline in an operational management and orchestration stack (that of EU 5Growth project) to analyze lifecycle management metrics in real-time, in this case the network service instantiation time related metrics. In our demonstration, a data connection module instance continuously monitors the NFV SO log files and sends the log changes to the data ingestion layer, where log files are temporarily stored to be fetched by Apache Spark jobs. After utilizing Spark jobs to cleanse the log files and to obtain the service instantiation times, the metrics are sent back to the data ingestion layer to be transferred to the Elasticsearch (ELK) stack for data indexing and visualization purposes. Furthermore, the statistical information of network service instantiation (in total and its components) of studied metrics inside the network can also be profiled via a separate data analysis layer connected to the ELK stack.

We experimentally demonstrate a 5G digital fronthaul network that relies on multi-adaptive bandwidth/bitrate variable transceivers (BVTs) and an autonomic software-defined networking (SDN) control system for partially-disaggregated wavelength division multiplexing (WDM)/space division multiplexing (SDM). Transmission of 256-QAM 760.32 MHz orthogonal frequency-division multiplexing (OFDM) radio signal is performed, with a total radio transmission capacity of 5.667 Gb/s. Digitized signal samples are carried as a 22.25 Gb/s digitized radio-over-fiber (DRoF) data stream and transmitted over a WDM/SDM infrastructure including 40-wavelength 100-GHz arrayed waveguide gratings (AWGs) and 19-core fiber. The autonomic SDN controller deploys a control loop for the multi-adaptive OFDM-based BVTs that monitors the per-subcarrier signal to noise ratio (SNR) and assigns the optimal constellation based on the actual signal degradation. An error vector magnitude (EVM) below the targeted 2.1% is achieved while setting up connections in less than 5 s.

5G networks aim to provide orchestration of services across multiple administrative domains through the concept of federation. In this paper, we are exploring the federation feature of a platform for 5G transport network of vertical services. Then we formulate the decision problem that directly impacts the revenue of 5G administrative domains, and we propose as solution a Q-learning algorithm. The simulation results show near optimum profit maximization and a well-trained Q-learning algorithm can outperform the intuitive “greedy” approach in a realistic scenario.

The composition of NFV-network services (NSs) helps 5G networks to provide the required flexibility and dynamicity to satisfy the needs of vertical industries. The mentioned capabilities are required not only at instantiation time but also during operation time to maintain the corresponding service level agreements in the presence of changing network conditions. In this demonstration, we present the capabilities of the 5Growth platform to handle the scaling of composite NSs under different situations. In particular, we show the scaling of a composite NS implying the update of the required deployed nested NS components and its interconnections but also how the scaling of a shared single NS entails updates on its associated composite NSs.

This paper presents the creation of Network Slices, which are composed of Network Services that are either composed of Cloud-native Network Functions or Virtual Network Functions. This is referred to as Hybrid Network Slice. In addition to describing this idea, this paper demonstrates the benefits of Hybrid Network Slices. This idea has been validated in an immersive-media pilot. An Hybrid Network Slice and a classical Network Slice deployments have been compared in order to validate the benefits from the hybrid case.

5G networks require flexibility, automation and programmability to satisfy the requirements of verticals industries. 5G-TRANSFORMER project proposes an SDN/NFV based network platform to enable this vision. Among its features, this platform allows the end-to-end deployment of parts of a network service (NS) in multiple administrative domains, which is known as network service federation (NSF). This feature increases network flexibility, opening the door to new business models. This paper complements our previous work by providing a detailed description of the 5G-TRANSFORMER NSF workflow, its interface and a profiling of the operations involved in the deployment of an NS between multiple administrative domains in a real experimental setup. Experimental results reveal i) that a federated NS can be deployed in the order of few minutes (less than 5 minutes), in line with the 5G target of reducing service setup to minutes, and ii) the impact of the NSF procedure in the deployment time is reduced when compared with the deployment of the same NS in a single administrative domain.

Current SDN controllers are monolithic applications that are run on dedicated servers and require specific protocols for synchronization among them. These SDN controllers do not provide the required flexibility to scale-out in case of a cloud-scale number of connectivity service requests. In this demonstration, We present the control of transport networks based on ONF Transport API using a cloud-native SDN controller based on micro-services. This demo provides insights on novel software implementations of transport network control technologies, such as container-based control architectures and standard interfaces. The proposed SDN controller components synchronize among them using gRPC protocol and a defined protocol buffer.

5G mobile networks need flexilibity, dynamicity and programmability to satisfy the needs of vertical industries. In such context, network services can be designed as integral units to be dynamically grouped among them to create tailored complex composite services or to allow the combination of services in a context of network slicing. In this demonstration, we present the service composition capabilities of the 5G-TRANSFORMER platform. In particular, we will show the instantiation of a composite network service using a previously instantiated service. When terminating the composite network service, the initial instantiated network service resumes its operation without disruption.

The future 5G transport networks are envisioned to support a variety of vertical services through network slicing and efficient orchestration over multiple administrative domains. In this paper, we propose an orchestrator architecture to support vertical services to meet their diverse resource and service requirements. We then present a system model for resource orchestration of transport networks as well as low-complexity algorithms that aim at minimizing service deployment cost and/or service latency. Importantly, the proposed model can work with any level of abstractions exposed by the underlying network or the federated domains depending on their representation of resources.

Future Internet of Things (IoT) will connect to the Internet billions of heterogeneous smart devices with the capacity of interacting with the environment. Therefore, the proposed solutions from an IoT networking perspective must take into account the scalability of IoT nodes as well as the operational cost of deploying the networking infrastructure. This will generate a huge volume of data, which poses a tremendous challenge both from the transport, and processing of information point of view. Moreover, security issues appear, due to the fact that untrusted IoT devices are interconnected towards the aggregation networks. In this paper, we propose the usage of a Software- Defined Networking (SDN) framework for introducing security in IoT gateways. An experimental validation of the framework is proposed, resulting in the enforcement of network security at the network edge.

This work extends NFV paradigm to transport networks, known as Transport NFV. This paper presents a detailed overview of the SDN/NFV services that are offered on top of the Cloud Computing platform and transport network of the ADRENALINE Testbed. On the one hand, we propose a generic architecture for SDN/NFV services deployed over multi-domain transport networks and distributed data centers. On the other hand, we present two use cases of possible NFV services: a virtual Path Computation Element (vPCE) and the deployment of virtual SDN controllers (vSDN) on top of virtualized transport networks.

A network hypervisor is demonstrated. The Virtual Network Controller dynamically deploys multi-tenant virtual networks on top of multi-technology transport networks. It provides an abstract view of each virtual network and enables its control through an independent SDN controller.

ITU-T Recommendation G.694.1 defines normative DWDM frequency grids, each being a reference set of values that correspond to allowed nominal central frequencies, obtained using a fixed channel spacing (e.g., 12.5 GHz, 25 GHz, 50 GHz or 100 GHz). This rigid, grid-based approach does not seem well adapted for data rates beyond 100 Gb/s, is particularly inefficient when a whole wavelength is assigned to a lower rate optical path, and is not flexible enough for multi-rate systems . Consequently, the next generation of optical networks will require a flexible, highly efficient and adaptive management of the optical spectrum, along with advanced optical modulation schemes that efficiently use allocated spectrum slots, and recent progress on optical network technology justifies research on both new optical transmission systems as well as the applicability of control and management frameworks to such networks. We design and deploy a GMPLS control plane for flexible optical networks with coherent optical orthogonal frequency-division multiplexing (O-OFDM) transmission; we detail its functional architecture, which combines a centralized entity that performs path routing and modulation assignment, with a distributed spectrum allocation. The centralized entity (i.e., a path computation element) uses pre-configured static path characterizations, based on exhaustive OFDM transmission simulations, when performing dynamic path computation in line with GMPLS constrained shortest path mechanisms. The distributed spectrum allocation assigns frequency ranges (slots) to connection requests, by using dynamic signaling procedures and applying slot assignment policies. We summarize the control plane protocol extensions involved in the main functional aspects: routing and topology dissemination, path computation, signaling and resource reservation. We experimentally validate and evaluate the integrated centralized PCE and GMPLS control plane in a control plane testbed, obtaining key performance indicators such as path setup latency and blocking probability for different frequency slot assignment policies.

In this paper we propose a dynamic multi-layer path computation algorithm which favors the re-usage of FA LSPs (Forwarding Adjacency Label Switched Path) and virtual links over a new LSP establishment. To evaluate the algorithm, we explore several approaches for the dynamic configuration of the Virtual Network Topology (VNT) within a multi-layer network (Connection - Oriented Ethernet transport over a Wavelength Switched Optical Network) controlled by a unified GMPLS control plane. Specifically, three VNT approaches are compared: the virtual approach, where some lambda FA LSPs are pre-computed but not established (i.e., virtual TE links), the semi-dynamic approach, where a specific set of lambda FA LSPs is pre-established and advertised as Ethernet TE links (i.e., FA TE links) and the dynamic approach, where there are neither pre-established nor pre-computed lambda FA LSPs.

To achieve cost reductions, simplified operations and flexible scalability in optical transport networks, a challenging approach is to replace the TDM aggregation and transport infrastructure, based on legacy SONET/SDH technology, by a new dynamic network infrastructure that delivers the high-bandwidth transport and deterministic performance of the optical circuit technology (i.e., WSON) along with the efficient aggregation and statistical multiplexing of a packet transport technology (e.g., MPLS-TP) to support IP and Ethernet services. This paper presents the architectural design, implementation and the performance evaluation of the forwarding capabilities of a MPLS-TP/PWE3 node with integrated 10Gbps tunable DWDM transponders to extend the GMPLS-enabled WSON transport network of the ADRENALINE testbed. The node has been implemented using commercial off-the-shelf hardware and open source software. An evaluation of the node is provided by means of analyzing the obtained throughput and CPU usage in two different evaluation scenarios with different traffic grooming strategies.

In Wavelength Switched Optical Networks (WSONs), sharing of protection wavelengths is an attractive strategy to increase survivability against failures. However, to guarantee an acceptable quality of transmission (QoT), both working and protection paths may need to undergo optical-electrical-optical (OEO) regeneration. With this aim, the placement of a limited number of regenerators is a cost-effective solution to guarantee QoT. In this paper, the concept of sharing the protection resources is extended to regenerators. Moreover, shared regenerators can be exploited for ensuring QoT as well as for providing wavelength conversion. The main objective of this paper is the study of different strategies for the selection of regenerators and wavelengths in WSON with a GMPLS-based distributed control plane. Simulation results show a trade-off between the strategies achieving a high wavelength sharing and those achieving a high regenerator sharing.

Sharing of protection resources is a cost-effective solution to guarantee survivability against failures. In wavelength-switched optical networks (WSON), in addition to sharing the wavelengths resources, wavelength converters (WCs) can also be shared by different optical connections, or lightpaths. This paper considers the problem of ensuring 100% survivability against single-link failures in a GMPLS-enabled optical network. RSVP-TE signalling protocol extensions for wavelength sharing under Shared Path Protection have been recently proposed for networks with wavelength continuity constraint. In this work, the performance of the proposed extensions is evaluated in a network with limited wavelength conversion. Furthermore, new extensions are proposed for an efficient WC sharing and their impact on the resource sharing is evaluated by simulations. Results indicate a trade-off between providing high wavelength sharing (i.e., low wavelength overbuild) and good WC sharing.

In this paper we propose novel wavelength assignment algorithms along with the required extensions to the current GMPLS RSVP-TE Label Set to efficiently encompass the Wavelength Continuity Constraint (WCC) and the Backup Sharing in GMPLS-enabled Wavelength-Routed Networks (WRN) with Shared Path Protection (SPP) recovery schemes. The focus of this paper is to experimentally validate in ADRENALINE testbed the proposed signalling-based solution for WCC in SPP recovery when GMPLS OSPF-TE routing protocol disseminates only aggregated link information at bandwidth level based on two approaches: Unreserved Bandwidth (UBw), as proposed in standard GMPLS, and Shamble Bandwidth (SRBw) through an extended GMPLS OSPF-TE protocol proposed by the authors. Experimental performance evaluation shows that without using full routing information (i.e., per-wavelength channel dissemination), the Resource Overbuild can be reduced up to 25% and the Blocking Probability up to 35%. Moreover, in function of the required performance, it is preferable to enhance the signalling with sharing information before adding it into the routing.

Generalized Multi-Protocol Label Switching (GMPLS) has become the protocol suite of choice for unified control plane implementation. However, its adoption is facing major challenges in terms of feasibility, performance, and gain when migrating from legacy packet over optical multi-layer networks driven by overlaid control planes. The ITEA TBONES project aims at tackling both objectives through the development of a platform including all the elements constituting such networks: network dimensioning, management plane and GMPLS control plane. This paper details the control plane architecture of the TBONES software platform, including its realization and applicability for multi-area networks, as well as the Testbed. Several experiments are described, both for validation and for demonstration of its capabilities and future usage.

This work focuses on the problem of dynamic provisioning of bidirectional connections in all-optical WDM networks. In absence of wavelength converters, each data path forming a bidirectional lightpath needs to use the same wavelength on every link of the entire route (i.e., wavelength continuity constraint). Though existing GMPLS framework faces up the bidirectional lightpath establishment, it fails with high probability under the WCC restriction. Therefore, a distributed and dynamic routing and wavelength assignment (DRWA) algorithm is proposed to perform bidirectional path computation taking into account global wavelength network utilization. By doing so, the algorithm is capable to meet efficiently with the WCC each particular connection request. The performance of the proposed algorithm is experimentally evaluated and compared in the ADRENALINE testbed using a traditional bandwidth-based algorithm as a benchmark. The results are expressed in terms of connection blocking and setup delay for different traffic loads.

Given the abundance and strategic importance of ring fiber-plants in metropolitan area networks, and the accelerating growth of Internet traffic, it is crucial to extend the existing IP-based GMPLS framework to provision dynamic WDM optical rings. Nevertheless the emerging GMPLS-based distributed lightpath establishment schemes for optical networks have been proposed and evaluated in mesh networks, but no one has been optimized for ring-based MAN. The objective of this paper is to propose a GMPLS-based signaling protocol for unidirectional rings which allows to have a global wavelength resource information without any routing protocol when provisioning bidirectional connections. Performance evaluation has been carried in ADRENALINE testbed, composed of Linux routers that emulate a GMPLS-based MAN.

Currently, GMPLS is being developed as a standard in order to introduce intelligence for next-generation ASON (automatically switched optical network) by means of a distributed optical control plane, allowing, among other functionalities, dynamic and real-time provisioning of high-bandwidth optical connections. The provisioning of bidirectional optical connections over unidirectional OADM (optical add drop multiplexer) rings using a distributed GMPLS-based control plane has not yet been considered in the existing literature. We present two enhancements to GMPLS signaling which allow a dynamically bidirectional lightpath setup to be requested over a unidirectional ring. These new distributed signaling schemes introduce a new concept of the label contention problem applied at wavelength sets. The paper presents and compares new strategies of label contention policies for wavelength sets, providing a significant improvement in performance evaluated in the GMPLS control plane of the EMPIRICO project testbed.