The growing demand for services and the rapid deployment of virtualized network functions (VNFs) pose significant challenges for achieving low-latency and energy-efficient orchestration in modern edge-core network infrastructures. To address these challenges, this study proposes a Digital Twin (DT)-empowered Deep Reinforcement Learning framework for intelligent VNF migration that jointly minimizes average end-to-end (E2E) delay and energy consumption. By formulating the VNF migration problem as a Markov Decision Process and utilizing the Advantage Actor-Critic model, the proposed framework enables adaptive and real-time migration decisions. A key innovation of the proposed framework is the integration of a DT module composed of a multi-task Variational Autoencoder and a multi-task Long Short-Term Memory network. This combination collectively simulates environment dynamics and generates high-quality synthetic experiences, significantly enhancing training efficiency and accelerating policy convergence. Simulation results demonstrate substantial performance gains, such as significant reductions in both average E2E delay and energy consumption, thereby establishing new benchmarks for intelligent VNF migration in edge-core networks.

The next generation networks offers significant potential to advance Intelligent Transportation Systems (ITS), particularly through the integration of Digital Twins (DTs). However, ensuring the uninterrupted operation of DTs through efficient computing resource management remains an open challenge. This paper introduces a distributed computing architecture that integrates DTs and Mobile Edge Computing (MEC) within a software-defined vehicular networking framework to enable intelligent, low-latency transportation services. A network aware scalable collaborative task provisioning algorithm is developed to train an autonomous agent, which is evaluated using a realistic connected autonomous vehicle (CAV) traffic simulation. The proposed framework significantly enhances the robustness and scalability of DT operations by reducing synchronization errors to as low as 7% while achieving up to 99.5% utilization of edge computing resources.

Six-generation mobile systems aim to support stringent end-to-end service-level agreements for diverse user applications simultaneously. This paper introduces novel multi-domain computation-aware resource slicing orchestration that jointly manages in-network communications, computation, and caching storage resources for multi-domain networking. It automatically programs wireless access, edge cloud, and regional/central cloud infrastructure to enable wireless-optical network virtualization. Specifically, a mobile virtual network operator's long-term profit maximization problem and two subproblems are formulated to slice wired and wireless infrastructure resources and assign user requests and contents to slices. Accordingly, a reinforcement learning-based slicing with greedy pre-caching is proposed, which automatically allocates in-network resources for dynamic wireless connectivity and supports real-time inferring with minimal user request signaling. Numerical results show that our solutions provide superior performance from both user and infrastructure perspectives, with 20% improved operator profits, 17% enhanced service provisioning rates, and 80% reduced delay when simultaneously serving augmented reality and large language model's quality demands. This innovation exploits a generalized rein-forcement learning approach to minimize the signaling overheads and computation complexity while agilely adapting to practical converged networks, thus benchmarking AI-driven multi-domain network slicing development.

The rapid evolution of Industry 4.0 has led to the emergence of smart factories, where multirobot system autonomously operates to enhance productivity, reduce operational costs, and improve system adaptability. However, maintaining reliable and efficient network operations in these dynamic and complex environments requires advanced automation mechanisms. This study presents a zero-touch network platform that integrates a hierarchical Open Radio Access Network (O-RAN) architecture, enabling the seamless incorporation of advanced machine learning algorithms and dynamic management of communication and computational resources, while ensuring uninterrupted connectivity with multirobot system. Leveraging this adaptability, the platform utilizes federated deep reinforcement learning (FedDRL) to enable distributed decision-making across multiple learning agents, facilitating the adaptive parameter reconfiguration of transmitters (i.e., multirobot system) to optimize long-term system throughput and transmission energy efficiency. Simulation results demonstrate that within the proposed O-RAN-enabled zero-touch network platform, FedDRL achieves a 12% increase in system throughput, a 32% improvement in normalized average transmission energy efficiency, and a 28% reduction in average transmission energy consumption compared to baseline methods such as independent DRL.

The need for prioritized multi-tenant quality of service (QoS) management in emerging mobile edge systems is particularly critical for high-throughput next generation networks. Current traffic engineering tools rely on network administrator driven, complex functions embedded in closed, proprietary infrastructures, which significantly restrict design flexibility, scalability, and adaptability. This study addresses these challenges by proposing a software-defined networking (SDN) based dynamic QoS provisioning scheme, powered by a digital twin (DT) of networks. By separating the control and data planes, the scheme enables automated traffic management through SDN programmability and data-driven decision-making. It incorporates few-shot learning to dynamically identify and prioritize multi-tenant network traffic utilizing flow statistics from SDN. The proposed QoS provisioning mechanism allocates sufficient bandwidth to high-priority flows while optimizing the remaining bandwidth for lower-priority traffic. Performance evaluations show that the model achieves up to 98% accuracy in identifying the priority of previously unseen traffic flows. Hardware-in-the-loop (HiL) simulations further validate the scheme's effectiveness in meeting multi-tenant QoS requirements, offering a robust and scalable solution for traffic prioritization in SDN based edge networks.

The integration of 5G terrestrial networks with Low Earth Orbit (LEO) satellites has the potential to provide seamless global connectivity and enhanced service quality, particularly in regions with limited terrestrial infrastructure such as rural areas. Furthermore, the incorporation of Multi-connectivity (MC) enables user equipment (UEs) to maintain simultaneous connections with both terrestrial 5G base stations and LEO satellites, improving system reliability. However, the high mobility of LEO satellites and the dynamic behavior of UEs present significant challenges, particularly in handover decision-making which can adversely impact system throughput, and quality of service (QoS). To address these challenges, we propose a novel deep reinforcement learning-based approach that integrates online Random Ensemble Mixture and Dual Experience Replay into a Dueling Double Deep Q-Network architecture. This proposed scheme intelligently optimizes handover decisions in MC-enabled terrestrial-satellite networks, improving decision accuracy in highly dynamic scenarios. Simulation results demonstrate substantial gains in system throughput, reduced system delay, average handover reduction, and increased transmission success probability, setting new performance benchmarks for integrated terrestrial-satellite networks while adhering to diverse QoS requirements.

The distributed and disaggregated architecture of next-generation (NextG) networks, including 6G has sparked growing interest in federated learning (FL) as a strategy for enabling privacy-preserving collaborative network traffic analysis at the edge. However, FL encounters significant challenges due to data heterogeneity driven by diverse data distributions across edge nodes, and the scarcity of labeled data further worsened by the time-intensive process of data labeling. Although a few studies have addressed these challenges in network traffic analysis tasks using Multi-Task Learning (MTL), existing approaches pre-dominantly focus on single-task FL, centralized model solutions and overlook the integration of MTL in NextG networks. To bridge this gap, we propose O-FedMTL, a novel framework that combines FL with MTL to enable cooperative traffic analysis within an Open Radio Access Network (O-RAN) environment in NextG networks. MTL enhances FL by mitigating the issues of data heterogeneity and labeled data scarcity through shared knowledge derived from multiple interconnected traffic analysis tasks, i.e., traffic classification, flow duration analysis, and bandwidth estimation. Additionally, MTL offers significant benefits by reducing energy consumption and computation costs at the edge through the simultaneous processing of these tasks within a single model. Extensive experimental results demonstrate that O-FedMTL achieves the target global accuracy for traffic classification, flow duration analysis, and bandwidth estimation with 20, 12, and 23 fewer global communication rounds, respectively, compared to the baseline federated averaging. Additionally, O-FedMTL reduces computation costs by 43% compared to the baseline-combined.

With the evolution of modern telecommunications, the fronthaul network has become an indispensable component of the infrastructure, particularly in the context of Cloud Ra-dio Access Networks. Nevertheless, fronthaul networks still face significant challenges such as power outages, particularly when a disaster, like an earthquake or severe weather event, occurs. Damage to power supply facilities may cause operational failures while communication is one of the most crucial needs in the disaster area to make rescue operations more effective. Specialized fibers that can deliver electrical power can help mitigate this problem. However, power losses may be extremely large over distances and there is no flexibility after the installation of the fibers. The network topology design that reduces capital and operational costs while satisfying power constraints is an important problem. In this paper, we focus on network topology design in an urban area by taking into account both fiber and power costs. We then propose integer linear programming and fast algorithms based on methods for single-facility location problems. The results demonstrate that multiple approaches help to achieve the optimal design, with our proposed method standing out due to its efficiency in finding feasible solutions, scalability, and ≈656x reduction in execution time.

In the realms of the internet of vehicles (IoV) and intelligent transportation systems (ITS), software defined vehicular networks (SDVN) and edge computing (EC) have emerged as promising technologies for enhancing road traffic efficiency. However, the increasing number of connected autonomous vehicles (CAVs) and EC-based applications presents multi-domain challenges such as inefficient traffic flow due to poor CAV coordination and flow-table overflow in SDVN from increased connectivity and limited ternary content addressable memory (TCAM) capacity. To address these, we focus on a data-driven approach using virtualization technologies like digital twin (DT) to leverage real-time data and simulations. We introduce a DT design and propose two data-driven solutions: a centralized decision support framework to improve traffic efficiency by reducing waiting times at roundabouts and an approach to minimize flow-table overflow and flow re-installation by optimizing flow-entry lifespan in SDVN. Simulation results show the decision support framework reduces average waiting times by 22% compared to human-driven vehicles, even with a CAV penetration rate of 40%. Additionally, the proposed optimization of flow-table space usage demonstrates a 50% reduction in flow-table space requirements, even with 100% penetration of connected vehicles.

C-band enabled Elastic optical networks (EONs) have been one of the most deployed optical network solutions in the world. However, as traffic demands continue to increase, capacity exhaustion is inevitable. There are two major technologies, namely, multiband elastic optical networks (MB-EONs) and space division multiplexed elastic optical networks (SDM-EONs) that can enhance capacity. Each technology offers a tradeoff between better capacity and deployment overhead which directly affects the network performance. Considering the different characteristics of these two technologies, we present our proposed strategy, Progressive Optics Deployment and Integration for Growing Yields (PRODIGY), to gradually migrate the current C-band EONs. PRODIGY uses various proactive measures, inspired by Swiss Cheese Model, to make the network robust for handling network traffic peaks and ensure that the service level agreement is met. We present a detailed comparison of our proposed strategy with customized baseline strategies, and demonstrate the superiority of our proposed approach.

We have demonstrated a 4×10 Gb/s RZ-DPSK wavelength multicasting using Raman amplification-based compressor with adjustable pulsewidth down to around 7.89 ps. Bit-error-rate at 10-9 of multicast signals are achieved with low power penalties.

We demonstrate a distributed multicarrier reusable network (DMRN) for regional and metro areas, based on dense wavelength-division multiplexing (DWDM) transmission with reconfigurable optical add/drop multiplexers (ROADMs). To eliminate the multiple distributed laser-diodes (LDs) at each access node in conventional ROADM networks, optical carriers generated by a centralized multicarrier light source (MCLS) are distributed to the access nodes, and they are used for node-to-node data transmission. The ROADM employed at each access node is used not only to “add” and “drop” data, but also to “drop” optical carriers. Moreover, to improve the wavelength utilization efficiency of the carriers distributed by the MCLS in the network, we proposed a technique called optical carrier regeneration (OCR), whereby the distributed carriers can be reused in each access node. This technique has a simple scheme and enables us to reuse the carriers that were already utilized for data transmission between prior source and destination nodes. In this work, we numerically analyze the scalability of our proposed DMRN in terms of the number of nodes, the span length, and the cascadability of the OCR. Moreover, we conduct a DMRN experiment using 10.7 Gb/s × 4 channels DWDM transmission and compare the transmission performances for various span lengths, for the first time. The results show that the DMRN will be useful for wide-area metro networks with high transmission performances.

Radio-over-Fiber (RoF) technology enables low-loss distribution of RF signals over optical fiber, which contributes much for spreading broadband wireless access services into in-building areas and outdoor dead-spots. However, current broadband wireless access systems employ OFDM schemes and suffer from nonlinear distortion of the RoF channel due to nonlinearity inherent in Electrical to Optical (E/O) conversion. In this paper, a new RoF transmission scheme employing Envelop Pulse-Width Modulation (EPWM) transmitter has been proposed to efficiently suppress the impact of RoF channel nonlinearity on OFDM signals. This idea is executed in a typical RoF channel that uses a Mach-Zehnder modulator as its E/O convertor. It is proved by both simulation and experiment that the EPWM-RoF scheme can achieve linear transmission of OFDM signal via nonlinear RoF channel.

This paper proposes a scheme to increase possible patterns of wavelength-conversion for an optical packet switch (OPS) with parametric wavelength converters (PWCs) by obtaining additional converted wavelengths using the existing resources. It is called a recursive parametric wavelength conversion (RPWC) scheme. A PWC uses a pump wavelength to define the original and converted wavelengths, called wavelength conversion pairs. None of conversion pairs from any PWCs can sometime support the original and available converted wavelength, although some wavelengths at the requested output fiber are available. Since the original wavelength is not converted, some packet losses may occur. Several conversion pairs are wasteful since they are not utilized. In RPWC scheme, unused conversion pairs are used to create additional conversion pairs. The OPS allows each wavelength to be converted using combination of unused conversion pairs using more than one PWCs, instead of using only a single PWC as in a conventional scheme. Numerical results via simulation show that the RPWC scheme achieves lower packet loss rate than the conventional scheme. To show the feasibility of the RPWC scheme, we develop a prototype of an optical switch with RPWC and demonstrate it in experiment.

In an optical packet switch (OPS), input fibers carry multiple wavelengths, which carry packets to one or more output fibers. As several wavelengths from different inputs could be destined to the same output fiber, one wavelength can be connected and the others remain disconnected, losing the carried packets. Because of the multiple wavelengths available at an output fiber, wavelength conversion in the OPS of the unconnected wavelengths into those available can increase the number of connections. A parametric wavelength converter (PWC) provides multi-channel wavelength conversion where wavelengths can be converted to another. A PWC uses a pump wavelength that can be flexibly chosen to define which wavelengths can be converted, defining the so-called wavelength conversion pairs. However, it is unknown which set of pump wavelengths, and therefore the set of connection pairs, should be selected to improve the OPS performance while minimizing the number of PWCs in the OPS. Therefore, this paper proposes a pump wavelength selection policy for an OPS that uses different pump wavelengths, one for each PWC, within an arbitrarily selected interval. This policy is called variety rich (VR) policy. This paper also introduces a non-wavelength blocking OPS (NWB-OPS) to make full use of PWCs. The switch performance is evaluated through computer simulation. The results show that the proposed policy with different pump wavelengths achieves the highest performance when compared to another of similar complexity. Furthermore, the performance study shows that small sizes of the interval to select a pump wavelength are more beneficial than larger ones.

This paper presents and experimentally demonstrates a multi-carrier distributed wavelength-division-multiplexing (WDM) ring network based on reconfigurable optical "drop-add-drop" multiplexers for regional and metro network applications. In the "drop-add-drop" network, optical carriers generated by a centralized multi-carrier light source (MCLS) are "dropped" at the source nodes and used for uplink transmission. Data are "added" to the network by external modulation of one or more carriers. Data are then "dropped" at the destination nodes. The reconfigurable optical add/drop multiplexer (ROADM) at each access node is not only used to "add" and "drop" data, but also to "drop" carriers, which eliminates the many distributed laser-diodes used in the conventional network. In this work, we successfully demonstrate, for the first time, a "drop-add-drop" network experiment offering 10 Gbit/s WDM transmission. Moreover, to dramatically improve the utilization efficiency of the carrier wavelengths distributed by the MCLS in the "drop-add-drop" network, we introduce the carrier wavelength reuse technique which sets the carrier extraction circuits in each access node. This technique enables us to reuse the carrier wavelengths that were already utilized for data transmission between prior source and destination nodes. To evaluate the effect of carrier wavelength reuse, we compare the blocking probabilities of the "drop-add-drop" networks with and without carrier wavelength reuse. The results show that wavelength reuse dramatically reduced the blocking probability. In addition, we numerically analyze the advantages of the "drop-add-drop" network over the conventional ROADM network in terms of network cost and power consumption.

This paper proposes an optical broadcast-and-select network architecture with centralized multi-carrier light source (C-MCLS). A large number of optical carriers/wavelengths generated by C-MCLS are distributed to all edge nodes (ENs), which select and modulate wavelengths to realize transmission. To utilize wavelength resources efficiently, we introduce a framework of wavelength allocation and selection (WAS). Wavelength allocation is performed at a wavelength control server, while wavelength selection is done at each EN according to wavelength allocation results. Both static and dynamic schemes are adopted for WAS and their implementations are shown. By using fixed or tunable band pass filter and periodical arrayed waveguide grating demultiplexer, wavelengths are selected and utilized by ENs in a static or dynamic manner. We evaluate network cost and performance of the proposed network. Cost analysis and numerical results show that it offers greatly reduced cost compared to the conventional one when the number of required access wavelengths at EN becomes large. We delineate its applicable areas through cost comparisons. Blocking probabilities of static and dynamic schemes are analyzed to evaluate network performance. Numerical results show that by choosing appropriate design parameters, the dynamic scheme offers about 25% increase in admissible offered load under the specified blocking probability, compared to the static scheme. This indicates that the dynamic scheme makes the proposed network more robust against traffic fluctuations.