Peer-to-peer disaggregated telemetry for autonomic machine-learning-driven transceiver operation

It is Software defined networking (SDN) is a new network approach which allows separation of the forwarding plane from the control plane in network architecture, that aids in the efficient control of data flow and gives network administrators a software- based method for managing and configuring the networks. In contrast to SDN, which provides a centralized controller, traditional network topologies require the individual configuration of each network device. Due to a centralized control functionality, SDN is increasingly used by businesses, telecom service providers, cloud providers, and many other users, as it encourages flexibility and scalability in the network while also assisting in streamlining operations. There is significant demand of SDN for effective and adaptable networks. SDN adoption has also been rising in the campuses and corporate sectors. This paper provides a concise overview of the Software Defined Networks (SDN) topic, which is essential for comprehending the fundamental principles of SDN in order to conduct research and implement SDN in many sectors. The initial focus is on the discussion of traditional networks, including their limitations, as well as the many issues encountered by industries. The topic of Software Defined Networks is explored, encompassing their necessity, historical background, architectural structure, interfaces, and models. Next, the many applications of Software-Defined Networking (SDN) are examined. Additionally, the various SDN case studies, vendors, setup issues with SDN, and current research fields were elucidated. The discussion focuses on the role of Software-Defined Networking (SDN) in Artificial Intelligence (AI), specifically addressing the issues associated with implementing SDN in AI.

New and emerging use cases, such as the interconnection of geographically distributed data centers (DCs), are drawing attention to the requirement for dynamic end-to-end service provisioning, spanning multiple and heterogeneous optical network domains. This heterogeneity is, not only due to the diverse data transmission and switching technologies, but also due to the different options of control plane techniques. In light of this, the problem of heterogeneous control plane interworking needs to be solved, and in particular, the solution must address the specific issues of multi-domain networks, such as limited domain topology visibility, given the scalability and confidentiality constraints. In this article, some of the recent activities regarding the Software-Defined Networking (SDN) orchestration are reviewed to address such a multi-domain control plane interworking problem. Specifically, three different models, including the single SDN controller model, multiple SDN controllers in mesh, and multiple SDN controllers in a hierarchical setting, are presented for the DC interconnection network with multiple SDN/ OpenFlow domains or multiple OpenFlow/ Generalized Multi-Protocol Label Switching (GMPLS) heterogeneous domains. In addition, two concrete implementations of the orchestration architectures are detailed, showing the overall feasibility and procedures of SDN orchestration for the end-to-end service provisioning in multi-domain data center optical networks.

Software Defined Networking (SDN) has recently emerged as a relevant network architecture enabling the direct programmability of forwarding functions and the effective abstraction of the underlying infrastructure. SDN has been successfully introduced in the context of packet switching (e.g., within data centers) and relevant effort has been recently devoted to extend the SDN architecture to efficiently operate on optical networks.

As the infrastructure of cloud computing and big data, data centers have been deployed widely. Then, how to make full use of the computing and storage resources in data centers will be the focus. Data center networks are considered important solution for the problem above, which include intra-data center and inter-data center networks. Both of them will depend on the optical networking due to its advantages, such as high bandwidth, low latency, and low energy consumption. Data center interconnected by flexi-grid optical networks is a promising scenario to allocate spectral resources for applications in a dynamic, tunable and efficient control manner. Compared with inter-data center, optical interconnect in intra-data center networks is a more pressing need and promising scenario to accommodate these applications in a dynamic, flexible and efficient manner. OpenFlow based Software Defined Networking (SDN) is considered as a good technology, which is very suitable for data center networks. This paper mainly focuses on the data center optical networks based on software defined networking (SDN), which can control the heterogeneous networks with unified resource interface. Architecture and experimental demonstration of OpenFlow-based optical interconnects in intra-data center Networks and OpenFlow-based flexi-grid optical networks for inter-data center are presented in the paper respectively. Some future works are listed finally.

Global optimization of optical network elements (NEs) is a potential solution to provide optical signal-to-noise ratio (OSNR) requirements for high spectrally-efficient (SE) modulation formats, essential in next-generation optical networks. In this context, software defined networking (SDN) is a suitable paradigm that allows for global monitoring and NEs actuation by decoupling data and control planes. By taking that into account, we review our recently proposed SDN dual-optimization application for EDFAs and WSS-based ROADMs which targets the optimization of the OSNR. In this work we detail the implementation of the application based on a state-machine approach. The application is tested using a SDN controller in a metropolitan optical network testbed. Experimental results show OSNR improvements of up to 10 dB for different application strategies and for different number of ROADM nodes in cascade.

Artificial intelligence based load balancing in SDN: A comprehensive survey

Software Defined Networking (SDN) revolutionizes network control by separating the control plane from the data plane. Although the latter improves SDN agility and scalability, it creates a security hole, particularly in a central control plane, leading to SDN environments becoming high-profile targets for advanced cybersecurity threats. Due to static and signature-based point-in-time behavior, traditional security methods are unable to keep up with modern attacks that are an anomaly to SDNs. Artificial Intelligence (AI) with its different applications and techniques, has the capability of detecting SDN cyber threats’ anomalies. This paper presents the results of a literature scoping exercise that used a total of 54 papers that looked at AI-driven anomaly detection in SDN. The findings showed that control theory, activity theory, and anomaly detection theory are three theoretical aspects that contribute to the topic of AI-driven anomaly detection in SDN. Furthermore, different machine learning algorithms give different results. In this regard, Random Forest (RF), Support Vector Machine (SVM), and Multi-Layer Perceptron would help in detecting threats of a familiar nature, while autoencoders and K-means can detect unfamiliar threats. While deep learning architectures such as Long Short-Term Memory (LSTM) networks and Convolutional Neural Networks (CNN) support low-latency anomaly detection while maintaining throughput and network stability. The findings could be the basis of providing a conceptual framework on how an intelligent, adaptive, and resilient SDN with real-time threat defense mechanisms could be designed, developed, and deployed.

The role of telecommunication and mobile networks changes from a medium that connects people through voice to a complex system that affects almost everything in our life. It is not just connecting people but also connecting things. It also acts as a medium for many services in almost all domains including health, education, security, and other domains. As telecom network grows in a complex way and our needs for a stable and high-quality service grow every day, there is a necessity to improve network and telecommunication services quality and ensure service availability 24/7. Nowadays we are also heading towards the industry 4.0 revolution, which revolutionizes maintenance, quality, and productivity. Network predictive maintenance is an important research field to take corrective actions that prevent network failure. In addition to that having autonomous network and telecommunication systems that could adapt to traffic changes and customer changing behavior becomes an important research field. Machine learning has always contributed to the implementation of predictive maintenance of telecom networks along with the autonomous self-adapted network. In this paper, we survey predictive maintenance in the telecommunication field and mobile networks. In addition to that, we survey network self-optimization through self-organizing networks, network slicing, and software-defined networks.

The rapid evolution of digital communication networks has underscored the necessity for intelligent, autonomous systems capable of adapting to dynamic data demands and complex operational environments. This presents a Model for Advancing Network Automation Through Software-Defined and Data-Driven Engineering Solutions, designed to enhance scalability, flexibility, and efficiency in modern network infrastructures. The proposed model integrates Software-Defined Networking (SDN), Network Function Virtualization (NFV), and Artificial Intelligence (AI)-driven analytics to enable real-time monitoring, automated configuration, and predictive optimization of network resources. By decoupling control and data planes, SDN provides centralized programmability and dynamic policy enforcement, while NFV introduces virtualized network functions that minimize hardware dependency and operational costs. The incorporation of Machine Learning (ML) algorithms facilitates data-driven decision-making, enabling automated fault detection, traffic prediction, and self-healing mechanisms that significantly improve network reliability and Quality of Service (QoS). The model also emphasizes data-centric engineering, leveraging big data analytics to continuously refine network performance and optimize bandwidth allocation based on traffic patterns and user behavior. Security is embedded through AI-based intrusion detection systems, encryption protocols, and policy-driven automation frameworks that ensure end-to-end protection. Furthermore, the model promotes energy-efficient operations and sustainable network management through intelligent resource scheduling and adaptive power utilization. Experimental evaluations and simulation results demonstrate substantial improvements in latency reduction, throughput enhancement, and fault tolerance compared to traditional network architectures. This contributes to the advancement of autonomous, self-optimizing digital ecosystems by bridging the gap between programmable networking and intelligent analytics. The proposed model provides a scalable foundation for next-generation applications, including 5G, IoT, cloud computing, and edge networks, ultimately driving the evolution toward fully automated, resilient, and adaptive communication infrastructures.

Today, there is no doubt that optical networks are the solution to the explosion of Internet traffic that two decades ago we only dreamed about. They offer high capacity with the use of Wavelength Division Multiplexing (WDM) techniques among others. However, this increase of available capacity can be betrayed by the high quantity of information that can be lost when a failure occurs because not only one, but several channels will then be interrupted. Efficient fault detection and location mechanisms are therefore needed. Our challenge is to identify and locate failures (single or multiple) at the physical layer of an optical network in the presence of some lost and/or false alarms. After briefly introducing optical networks and the multiplexing techniques that can be used, we study the most common components and their most usual failures. We propose a classification of all the network components based on their behaviour when failures occur. This classification gives an abstract model of the optical network, which is appropriate for developing algorithms to locate faulty elements. Two algorithms that solve the fault location problem are proposed. Both algorithms cope with existence of false and missing alarms when locating single and multiple failures. The problem of locating multiple failures already in the absence of false or missing alarms, has been shown to be NP-complete. The first algorithm, which is called Alarm Filtering Algorithm (AFA) is based on the combination of two approaches: forward and backward. The forward approach returns for each network element, their domain, which is the set of network elements that will send an alarm when the considered element fails. The backward approach returns the set of elements that are directly related to the received alarms. In this approach, the alarms that are considered to provide redundant information, are discarded. The combination of the results given by both approaches allows the location of multiple failures, given an allowed number of false and missing alarms. However, this algorithm does not minimize the complexity when new alarms are received. Hence, a second algorithm, which is called Fault Location Algorithm (FLA), is proposed. The FLA concentrates the complexity in ,a pre-computation phase, so that when new alarms are received, the result of the algorithm is rapidly displayed. The FLA algorithm is based on the construction of a binary tree that realizes a non linear error correcting code. The FLA has also been extended to locate soft failures in addition to hard failures. Hard failures are unexpected failures, whereas soft failures are progressive failures due to equipment aging, misalignments or external factors such as temperature or pressure. Both algorithms are compared on some simulated networks using different network topologies and failure cases. The comparison has also be done on the basis of their worst case complexity. Some conclusions indication with which settings each algorithm perform the best, were obtained.

It is known that the design of the control plane (CP) is vital in the network planning for software-defined networking (SDN), while to improve throughput and enlarge geographical coverage, optical networks are commonly used as the physical infrastructure of SDN. However, physical-layer attacks can disrupt the operation of an optical network and thus complicate the CP design. In this work, we consider planned physical-layer attacks when designing the CP of an SDN that uses an optical network as its physical infrastructure. We first show that the CP design problem should be modeled as a bilevel optimization, where the network planner designs the CP with the minimum vulnerability to physical-layer attacks (i.e., the upper-level optimization), while the attacker plans and launches attacks to disrupt the designed CP (i.e., the lower-level optimization). Then, the bilevel mode is transformed into a mixed integer linear programming (MILP) model, which can solve the CP design problem exactly. We also propose a heuristic to tackle the problem time-efficiently.

The advancement of modern multimedia and data-intensive classes of applications demands the development of hardware that delivers better performance. Due to the evolution of 5G, Edge-Computing, the Internet of Things, Software-Defined networks, etc., the data produced by the devices such as sensors are increasing. A software-Defined network is a powerful paradigm that is capable of automating networking and cloud computing. Software-Defined Network has controllers, devices, and applications which produce a huge amount of data. The processing of data inside the device as well as between the devices needs a better hardware architecture with more cores to ensure speedy performance. The System-on-Chip approach alone will not be capable to handle this dense core comprised of hardware. We have to blend Network-on-Chip along with System-on-Chip to increase the potential to include more cores capable to handle more threads. Artificial Intelligence, a key enabler in next-generation devices is capable of producing a better architecture design with optimized performance. In this paper, we are discussing and endeavouring how System-on-Chip, Network-on-Chip, Software-Defined Networks, and Artificial Intelligence can be physically, logically, and contextually incorporated to deliver improved computation and networking outcomes.

In this paper, we describe the integration of Software Defined Networking (SDN) in the mobile backhaul as a disruptive approach to streamline the transport network. In this work we leverage SDN to optimize the mobile backhaul transport by removing all mobile specific tunnelling and replace it with more efficient MPLS or Carrier Grade Ethernet deployed either over electrical or optical networks. The paper also presents the testbed with complete end to end system including off the shelf base stations, SDN enabled mobile backhaul switches and virtualize network elements (i.e. Mobility Management Entity (MME), Serving/Packet Gateway (S/P-GW)) running on the cloud. This testbed is currently accepted as European Telecommunication Standards (ETSI) Proof of Concept and the results are used to describe the benefits for operators and end users. Moreover, an initial design of services based on the proposed virtualized mobile network architecture is proposed. The results of the testbed show the benefits for mobile operators in terms of Capital Expenditure (CAPEX) and Operational Expenditure (OPEX) savings but more importantly the development of services that benefit from optimal usage of resources.

Currently, a wide demand for the implementation and use of various cloud solutions is a modern trend and the driving force behind the development of network technologies. The growth of cloud application services delivered through data centers with varying network traffic needs demonstrates the limitations of traditional routing and load balancing methods. The combination of the advantages of software defined networks (SDN) technology and artificial intelligence (AI) methods ensures efficient management and operation of computer network resources. The paper proposes an approach to neural network multipath routing in SDN based on a genetic algorithm. The architecture and model of an artificial neural network has been developed to solve the problem of multipath routing in the SDN, which is able to predict the shortest paths based on the metrics of communication links. To optimize the hyperparameters of the neural network model, it is proposed to use a modified genetic algorithm. A visual software system SDNLoadBalancer has been developed and an experimental SDN topology has been designed, which makes it possible to study in detail the processes of neural network multipath routing in SDN based on the proposed approach. The obtained results show that the proposed neural network model has the ability to predict routes with high accuracy in real time, which makes it possible to implement various load balancing schemes in order to increase performance of SDN.

Optical networks, which were once considered to be static, are now becoming more flexible and dynamic. Software-defined networking (SDN) has an essential role in the provisioning and management of these networks. There is also a requirement to describe the heterogeneous network resources in a unified manner to the SDN controller. In this paper, we develop a novel semantic framework called Optical Network Description Language (ONDL) which is used to describe entities in the optical network. We show its applicability by emulating a physical layer topology and its control using the developed SDN controller based on ONDL. For any network event, we obtain the response time as <100 ms and observe that the total time to configure a device is dependent upon the individual node configuration time.