The growing complexity of cloud-native, edge, and IoT infrastructures has made manual configuration, fault remediation, and lifecycle management increasingly unsustainable. Traditional automation techniques—such as rule-based logic or bespoke machine learning pipelines—struggle with adaptability and explainability in dynamic environments. Recent advances in Large Language Models (LLMs), however, have introduced new opportunities for autonomous, intent-driven infrastructure control. In this work, we present a closed-loop framework that integrates LLM agents for automated Root Cause Analysis (RCA) and mitigation of faults within cloud-edge and IoT systems. When SLA violations are detected, the agent identifies likely root causes and selects corrective actions—such as pod rescheduling, scaling, or configuration updates—executed via a Model Context Protocol (MCP) server exposing management tool functionalities through an API. This RCA-plus-mitigation loop enables fault handling that is both explainable and adaptive. We evaluate our system on a cluster running synthetic IoT workloads under emulated stressors using a reproducible benchmarking setup. Results show that the agent identifies SLA violations with 52.9% accuracy and mitigates 70.7% of them successfully. Notably, the agent incorporates validation steps to ensure system stability after interventions. These findings highlight the feasibility of LLMs for real-time infrastructure healing and their potential role in future AIOps workflows.

Fully autonomous, zero-touch systems emphasizing on energy efficiency, high reliability, and ultra-low latency will be possible with the introduction of 6G networks. But with more devices and services, energy usage is expected to rise, necessitating sustainable solutions. A Decision Engine (DE) based on Hierarchical Reinforcement Learning (HRL) is presented in this research to improve the deployment of Service Function Chains (SFCs) based on Cloud-Native Functions (CNF) in dynamic contexts. The goal of the framework is to lower energy consumption while improving scalability and flexibility in the cloud, far-edge, and edge domains. By simulating actual 6G situations, we demonstrate that the HRL-based DE improves resource allocation, reduces latency by 80%, and considerably reduces energy usage by 60% compared to the flat RL. By assisting in self-optimizing network management, our method presents a viable route to intelligent, sustainable 6G networks.

In the realm of modern telecommunication ecosystems, resource-aware management is critical in ensuring quality of service (QoS) while enabling proactive, zero-touch operations. Furthermore, recent advances in software defined networking (SDN) have significantly contributed to beyond 5G (B5G) network optimization, offering innovative solutions to tackle their growing complexity. Moreover, the inherently non-convex nature of service provisioning challenges presents critical obstacles to delivering a seamless user experience. This paper introduces a novel framework that integrates Dueling Double Deep Q-Networks (Dueling DDQN) within a cloud-native edge infrastructure, enabling zero-touch service scaling. By harnessing the advanced capabilities of deep reinforcement learning (DRL), the proposed method autonomously identifies and addresses scaling demands, thereby improving overall network capabilities. The effectiveness of our approach is validated through experiments conducted on a multi-node Kubernetes testbed, demonstrating its ability to alleviate performance bottlenecks in distributed environments.

The emergence of the 6thgeneration (6G) of networks introduces new challenges in meeting users’ unparalleled demands for high performance and reliability. To meet these escalating requirements, the efficient orchestration of the network resources becomes essential. In this work, we propose a Spatio-Temporal Graph Neural Network (STGNN)-based approach for predicting the users’ resource demand targeting to enhance the resource management and in parallel to guarantee Service Level Agreement (SLA) compliance. Our approach mitigates key challenges in resource orchestration, such as under-provisioning, under-utilization, and user blocking by integrating relational and temporal information into the prediction process. The evaluation of the proposed framework incorporating these predictions demonstrates a relative reduction of up to 34% in SLA violations through proactive resource allocation. These results underscore the potential of STGNNs to enable intelligent resource management in next-generation mobile networks.

Advancement of edge and far-edge computing, driven by the increasing demand for real-time, data-intensive applications, has heightened the need for reliable and efficient resource management in volatile environments. This paper introduces GTMixer, a deep learning architecture tailored for predicting resource usage in volatile edge computing scenarios. GTMixer utilizes a Dynamic Temporal Graph (DTG) to capture the evolving interdependencies in workload exchanges across edge and far-edge nodes. By processing snapshots of this graph, GTMixer identifies patterns in resource utilization, even in the absence of historical CPU data. Our contributions include: (1) the creation of a DTG that reflects migration patterns and resource utilization among nodes; (2) the development of the GTMixer model, which integrates feature mixing with graph neural networks for improved predictive accuracy; and (3) empirically evaluating GTMixer against state-of-the-art models using a modified version of the Alibaba 2021 traces dataset that accounts for volatility. Our results demonstrate that GTMixer not only effectively anticipates resource requirements in unpredictable Multi-access Edge Computing (MEC) scenarios but also significantly outperforms current state-of-the-art models in terms of efficiency, showcasing its potential to enhance the reliability and performance of edge computing systems crucial for nextgeneration network technologies and applications.

Although the next generations of wireless networks are anticipated to be enabled by Artificial Intelligence (AI), the development of practical, scalable, and efficient system models and network orchestration strategies remains a significant open challenge that requires thorough investigation. In this research work, we study a system model including two parts of AI-driven management and orchestration and AI-enabled infrastructures consisting of multiple automated domains. While the network orchestration manages the autonomous domains, the intra-domain resource allocation and network slicing are also automated and AI-enabled. Consequently, we propose an efficient and scalable strategy for automating network edge domains. This approach reduces the complexity and, thus, increases the scalability and energy efficiency by implementing one Double Deep Q-Learningbased controller in each domain that can perform network slicing for multiple service types. One agent provides dynamic network services by utilizing unsupervised discovered assignment skills in an offline phase. We implement the domain controller with multiple algorithms to find the best performance and efficiency. Finally, we compare the reliability of the algorithms, and we select the algorithm that offers the best trade-off between complexity and performance.

The 6thgeneration (6G) of mobile networks are anticipated to provide significant benefits such as high data-rates, low-latency and seamless connectivity to a huge number of users. The realization of these networking potentialities is anchored in the innovative design and development of the networking operations with purpose to deal with complex environments in a efficient manner. In particular, principal networking operations such as the resource management will be mainly conducted with the aid of the Deep Learning (DL) models. The DL models are characterized by their efficiency in dealing with intense tasks but they lack of transparency and trust. For this reason, we propose a framework that incorporates both a methodology for uncertainty quantification (UQ) of the DL models and a trafficengineering model targeting on improving the trustworthiness in the resource allocation procedures. It is shown that the exploitation of trustworthy DL models significantly improves both the system's resource utilization and the system's service provisioning capability to the users.

This research introduces an advanced Explainable Artificial Intelligence (XAI) framework designed to elucidate the decision-making processes of Deep Reinforcement Learning (DRL) agents in O-RAN architectures. By offering networkoriented explanations, the proposed scheme addresses the critical challenge of understanding and optimizing the control actions of DRL agents for resource management and allocation. Traditional methods, both model-agnostic and model-specific approaches, fail to address the unique challenges presented by XAI in the dynamic and complex environment of RAN slicing. This paper transcends these limitations by incorporating intent-based action steering, allowing for precise embedding and configuration across various operational timescales. This is particularly evident in its integration with xAPP and rAPP sitting at near-real-time and non-real-time RIC, respectively, enhancing the system's adaptability and performance. Our findings demonstrate the framework's significant impact on improving Key Performance Indicator (KPI)-based rewards, facilitated by the ability to make informed multimodal decisions involving multiple control parameters by a DRL agent. Thus, our work marks a significant step forward in the practical application and effectiveness of XAI in optimizing O-RAN resource management strategies.

With the advent of beyond 5G applications, the execution of computationally intense tasks moves further closer to the network edge. Alongside the capabilities of a Multi-Access Edge Computing (MEC), smart decision-making considering sustainability aspects has become achievable. In this paper, a resource management technique utilizing Reinforcement Learning (RL) at the MEC is presented in order to promote power efficient solutions. CPU resources at the MEC are managed and distributed to several network services for their individual disposal. A direct relation between the CPU resources and power consumption at the MEC is proposed further establishing the need for efficient resource handling. A Soft-Actor Critic (SAC) approach is leveraged to learn the patterns for intelligent resource allocation minimizing the power expenditure. Further, two baseline algorithms, the Knapsack method and the proportional resource allocation scheme, are implemented to prove the dominance of the proposed RL-based algorithm. The results confirm that in the SAC-based RL implementation, the power consumption at the MEC server is lower compared to the two baseline algorithms. The promising results pave way for the deployment of RL-based algorithms for efficient performance, thus promoting green technologies at the MEC.

Next-generation telecommunication networks provide vast amounts of Multivariate Time Series monitoring data during their operation. Using these data to automatically detect anomalous system behavior is of critical importance to operators and Internet Service Providers, in order to analyze the network state and take mitigating actions to resolve possible faults. For the output of this Anomaly Detection process to be trusted, comprehensive and accurate explanations must be provided. These explanations are essential to make critical decisions that affect the network state. In this paper, we present a framework that utilizes reconstruction-based Deep Neural Network models to perform the anomaly detection task, while also using an explainability module that provides explanations of the results. The quality of the explanations is quantified through the use of interpretation labels. In addition, we provide a comparative analysis, performing Anomaly Detection on three different open-source datasets from the mobile networking domain, while testing multiple Deep Learning models and comparing their performance. The presented results show that our framework accurately detects various types of anomalies that can be present in mobile networks, while offering the flexibility to identify and select the model that offers the best performance for each dataset, in both the Anomaly Detection and explainability tasks.

Cell- Free (CF) massive Multiple- Input- Multiple-Output (mMIMO) has recently gained significant research attention as a promising technology for future wireless networks. Since the conventional CF mMIMO has been considered unscalable and impractical, user-centric CF systems were proposed to improve its flexibility. While Access Point (AP) clustering has been studied in many research works as a challenging task in Radio Access Network (RAN), the limitations of the connecting links to the servers in a real scenario have not been taken into account. In this work, we consider the innovative combination of Open RAN (ORAN) and CF-RAN architecture that aims to improve the CF network limitations related to the connecting links. Then, we propose two control loops for ORAN Radio Unit (ORU) clustering and ORAN Distributed Unit (O-DU) assignment procedures that are conducted by Reinforcement Learning (RL) agents. Numerical results show that the proposed approach can successfully provide the user's requirements while the network is balanced.

This paper addresses how to handle slice resources for 6G networks at different time scales in an architecture based on an open radio access network (ORAN). The proposed solution includes artificial intelligence (AI) at the edge of the network and applies two control-level loops to obtain optimal performance compared to other techniques. The ORAN facilitates programmable network architectures to support such multi-time scale management using AI approaches. The proposed algorithms analyze the maximum utilization of resources from slice performance to take decisions at the inter-slice level. Inter-slice intelligent agents work at a non-real-time level to reconfigure resources within various slices. Further than meeting the slice requirements, the intra-slice objective must also include the minimization of maximum resource utilization. This enables smart utilization of the resources within each slice without affecting slice performance. Here, each xApp that is an intra-slice agent aims at meeting the optimal quality of service (QoS) of the users, but at the same time, some inter-slice objectives should be included to coordinate intra- and inter-slice agents. This is done without penalizing the main intra-slice objective. All intelligent agents use deep reinforcement learning (DRL) algorithms to meet their objectives. We have presented results for enhanced mobile broadband (eMBB), ultra-reliable low latency (URLLC), and massive machine type communication (mMTC) slice categories.

Network Virtualization (NV) has proved a promising technology that allows multiple heterogeneous Virtual Networks (VNs) to operate simultaneously on the same infrastructure. Dynamic Virtual Network Embedding (NVE) has emerged as an enabler of elasticity and scalability in the VN deployment and resource allocation of the physical infrastructure. However, the key challenge in realizing NV in a sustainable way is how to dynamically embed VNs efficiently into physical network, which is defined as the VNE problem. To address this challenge, this paper proposes an approach that leverages Multi-Agent Reinforcement Learning (MARL) to solve the dynamic VNE of VNs for 5G/6G communication systems. The proposed framework consists of multiple horizontally distributed RL agents that co-operate to devise temporally dynamic VNE placements. The key contributions of this work are introducing a novel dynamic VNE orchestration framework for multi-domain networks based on Distributed RL, providing a scalable VNE framework targeted to Ultra-Reliable Low-Latency Communication (URLLC) services, evaluating and comparing the proposed algorithm with existing solutions in the state of the art. The paper concludes that there is a significant improvement in latency of 144.191% when compared to the baselines.

As commercial 5G networks become commercially available and 6G looms in the horizon, the adoption of this new technologies depends on the way current and yet-to-come vertical industries put them to use. This accelerates the process of adoption by the end users, which are the final consumers of these technologies. 5G networks have multiple use cases associated with its new features, being the Ultra-Reliable Low Latency Communications (URLLC) use case one of the most instrumental for verticals such as remote teleoperation, factory automation and autonomous driving. In this paper, we design a general purpose and low-cost end-to-end (E2E) 5G/Beyond-5G Experimental Platform for URLLC applications, supporting Platform-as-a-Service (PaaS) with Artificial Intelligence (AI) and Machine Learning (ML) capabilities for online data analytics and automated decision making. The experimental evaluation of our platform demonstrates an average one-way latency on the DL and UL as low as 4.1 ms and 6.60 ms, respectively, 13.8 ms of end-to-end latency (E2EL), and a 31.7 ms E2EL between UEs for data frames of a teleoperation Web App with video feedback, demonstrating the capability of our platform in relation to other state-of-the-art testbeds for URLLC applications.

Integrated Access and Backhaul (IAB) has been recently proposed by 3GPP to enable network operators to deploy fifth generation (5G) mobile networks with reduced costs. In this paper, we propose to use IAB to build a dynamic wireless backhaul network capable to provide additional capacity to those Base Stations (BS) experiencing congestion momentarily. As the mobile traffic demand varies across time and space, and the number of slice combinations deployed in a BS can be prohibitively high, we propose to use Deep Reinforcement Learning (DRL) to select, from a set of candidate BSs, the one that can provide backhaul capacity for each of the slices deployed in a congested BS. Our results show that a Double Deep Q-Network (DDQN) agent using a fully connected neural network and the Rectified Linear Unit (ReLU) activation function with only one hidden layer is capable to perform the BS selection task successfully, without any failure during the test phase, after being trained for around 20 episodes.

The evolution of the telecommunication networks towards their 6th generation has emerged new research directions in order to deal with the challenges that are posed by the high-complexity of the new infrastructure. In this new era, the structuring of the network management infrastructure is a complex endeavor that should efficiently control both communication and computational resources. In this paper, we present a set of strategies for managing the computational load in a cell-free-based, converged optical-wireless 6G network. The proposed approaches target to control the computational load of a cell-free network, by either compressing the load by considering load-thresholds, or offloading the load to the SDN controller of the fixed network. Both strategies are mathematically formulated by considering traffic-engineering formulas, while their performance is evaluated by comparing analytical results with corresponding results from a baseline scenario, where no load control strategies are applied. Moreover, the proposed analysis can be applied in order to determine the capacity of the network controllers that is required in order to guarantee pre-determined Quality of Service requirements.

Managing network slices in 5G networks and in communication technologies Beyond-5G (B5G) requires intelligent mechanisms to ensure users’ service access and to maximize the utility and efficiency of the network’s physical resources. To achieve this, we propose a mechanism based on Reinforcement Learning (RL) for the Admission Control (AC) of User Service Requests (USRs) into network slices through dynamic bandwidth (BW) reallocation. Our approach admits, delays or rejects USRs into service depending on the BW of the slice and the utility this generates for the infrastructure provider (InP). This approach achieves very low USR rejection rates (RRs) with very high resource efficiency, even when peak traffic loads considerably exceed the BW capacity causing resource scarcity scenarios. When compared against a static BW allocation mechanism, our approach achieves RRs that are a fraction (0,33) of those achieved by static (smaller RRs are better), with 33,2x less resource overallocation, significantly achieving a very high resource efficiency.

The evolution of Software-Defined Networking (SDN) and Network Function Virtualization (NFV) in the telecommunications industry have intensified the issues of network management at large scales. Dynamic service orchestration and adaptive resource allocation became a necessity for network operators to manage the rapid growth of users and data-intensive applications. The impact of network automation on energy consumption and overall operating costs is often overlooked. Guaranteeing strict performance constraints of Ultra-Reliable Low Latency Communication (URLLC) services while enhancing energy efficiency is one of the major critical problems of future communication networks, given the urgency to reduce carbon emissions and energy consumption. In this work, we study the problem of zero-touch Service Function Chain (SFC) orchestration for multi-domain networks, targeting the latency reduction of URLLC services while improving energy efficiency for beyond-5G networks. Specifically, we propose SCHE2MA, a Service CHain Energy-Efficient Management framework based on distributed Reinforcement Learning (RL), that can intelligently deploy SFCs with shared VNFs per se into a multi-domain network. Finally, we evaluate SCHE2MA through model validation and simulation while demonstrating its ability to jointly reduce average service latency by 103.4% and energy consumption by 17.1% compared to a centralized RL solution.

Service providers highly rely on network softwarization for addressing the demanding use cases of the 5G verticals. In order for the 5G vertical scenarios to be seamlessly executed, where the coverage should be provided in large areas, the required infrastructure capacity and availability can only be supported by cross-operator cooperation. In this context, we employ a novel federated core-edge 5G architecture, where 5G vertical services are represented as service function chains (SFCs) and can be offered on the infrastructure owned either by the local operator or a foreign one. Furthermore, we propose two online SFC placement methods that aim to efficiently place the SFCs across the architecture, taking into account the deployment cost. A simulation of the optimal solution, based on an integer linear programming (ILP) approach, is evaluated against a hop-based heuristic placement method that runs on our experimental 5G platform. Our experimental results demonstrate that the optimized placement produces the most cost-effective solution. Our heuristic algorithm introduces a near-optimal solution, but with a higher deployment cost. Compared with the ILP approach, the heuristic solution provides lower complexity and execution time. Finally, in order to further augment the placement techniques, we propose enhanced lifecycle management methods that use live migration and scaling actions to further decrease the execution cost and adapt to real-time incoming traffic, respectively.

As the demand for Network Function Virtualization accelerates, service providers are expected to advance the way they manage and orchestrate their network services to offer lower latency services to their future users. Modern services require complex data flows between Virtual Network Functions, placed in separate network domains, risking an increase in latency that compromises the offered latency constraints. This shift requires high levels of automation to deal with the scale and load of future networks. In this paper, we formulate the Service Function Chaining (SFC) placement problem and then we tackle it by introducing SCHEMA, a Distributed Reinforcement Learning (RL) algorithm that performs complex SFC orchestration for low latency services. We combine multiple RL agents with a Bidding Mechanism to enable scalability on multi-domain networks. Finally, we use a simulation model to evaluate SCHEMA, and we demonstrate its ability to obtain a 60.54% reduction of average service latency when compared to a centralised RL solution.

Future mobile networks will offer high data rates based on high-capacity fronthaul. Current fronthaul design has two main components that communicate via the common public radio interface and fiber links, i.e., remote units (RUs) that implement simple signal processing and centralized baseband units (CBBUs) in high power-consuming data centers that perform complex network functions. Various functional splits between CBBUs and RUs are feasible, inducing trade-offs between centralization gains and bandwidth demands. This design lacks in capacity and flexibility, motivating the use of converged fiber-wireless (Fi-Wi) fronthaul with high-bandwidth fiber and millimeter-wave links, and splits that move functionalities to RUs reducing the delay demands. Further flexibility is offered by analog radio-over-fiber fronthaul that supports dynamic functional splitting via software-defined networking (SDN). Ensuring acceptable delay for all RUs, i.e., minimizing fronthaul grade-of-service (GoS), requires selection of CBBUs, channel bandwidth and functional splits of RUs. The split type affects fronthaul power consumption determining which fronthaul components are active and their processing power. Using a simulated annealing-based dynamic fronthaul resource allocation (DFRA) scheme, we jointly optimize GoS and power consumption in a novel SDN Fi-Wi fronthaul. Our results show that DFRA minimizes GoS and power consumption for all load levels outperforming baseline approaches.

Network Slicing (NS) represents a key technology enabler for advanced connectivity and data processing tailored to customers' specific requirements. While significant progress has already been achieved for Core NS, Radio Access Network (RAN) slicing still presents limitations in terms of sharing infrastructure, Service Level Agreement (SLA) guarantees, isolation, resource scheduling and allocation. In this context, this paper firstly introduces a novel slices configuration framework for the 5G New Radio (5G NR) infrastructure able to dynamically migrates the radio resources among the slices, while preserving the Quality of Service (QoS) of the served users. Our solution is illustrated in detail and tested on top of a real case 5G scenario, using a software-based simulator. Finally, this paper investigates the flexibility, scalability, and real-time properties of the proposed method, as required in the future 5G cloud-based architectures.

The upcoming fifth generation (5G) mobile communications urge software-defined networks (SDNs) and network function virtualization (NFV) to join forces with the multiaccess edge computing (MEC) cause. Thus, reduced latency and increased capacity at the edge of the network can be achieved, to satisfy the requirements of the Internet of Things (IoT) ecosystem. If not properly orchestrated, the flexibility of the virtual network functions (VNFs) incorporation, in terms of deployment and lifecycle management, may cause serious issues in the NFV scheme. As the service level agreements (SLAs) of the 5G applications compete in an environment with traffic variations and VNF placement options with diverse computing or networking resources, an online placement approach is needed. In this article, we discuss the VNF lifecycle management challenges that arise from such heterogeneous architecture, in terms of VNF onboarding and scheduling. In particular, we enhance the intelligence of the NFV orchestrator (NFVO) by providing: 1) a latency-based embedding mechanism, where the VNFs are initially allocated to the appropriate tier and 2) an online scheduling algorithm, where the VNFs are instantiated, scaled, migrated, and destroyed based on the actual traffic. Finally, we design and implement an MEC-enabled 5G platform to evaluate our proposed mechanisms in real-life scenarios. The experimental results demonstrate that our proposed scheme maximizes the number of served users in the system by taking advantage of the online allocation of edge and core resources, without violating the application SLAs.

As industries are under pressure for shorter business and product life cycles, there is an extensive effort from the research community for novel and profitable automation processes. This effort has given rise to the fifth-generation (5G) Tactile Internet, which is characterized by extremely low latency communication in combination with high availability, reliability, and security. In this paper, we discuss the key technologies to support the Tactile Internet characteristics in industrial environments and then showcase the implementation of a novel 5G network function virtualization enabled experimental platform. Given that ultra-reliable low-latency communication is crucial for the manufacturing process, we demonstrate that, in our setup, submillisecond end-to-end communication is attainable, proving the suitability of our platform for Tactile Internet industrial applications.

The 5G networks are expected to satisfy diverse use cases and business models with significant advancements in terms of capacity, reliability, and latency. The allocation and provisioning of network resources pose a challenge for this novel architecture to guarantee higher flexibility and quality of service. As a potential enabler, network slicing was proposed as an innovative approach for the control of the network resources. Although a static slicing approach can be suitable for the transport and core network, the stochastic behavior of the wireless channel requires fast and secure slicing techniques for resource allocation. In this paper, we propose a dynamic slicing approach for the radio access network, where the network resources are carefully assigned to guarantee the service level agreements and increase the number of served users. To prove the performance of our approach, we implemented a fronthaul testbed to emphasize the strength of our method in terms of throughput and resource utilization, compared to static slicing.

In a fluctuating mobile environment where operators have to confront the ever increasing demands of their subscribers, insufficient spectrum poses capacity limitations. This is more evident in the downlink (DL) direction, since DL resources are over-utilized compared to the uplink (UL) ones as a result of asymmetry in the generated traffic and intense interference. In this framework, we propose the creation of Device-to-Device (D2D) based clusters of users where intra-cluster communication will be achieved over UL resources. The minimization of the required resources (equivalent to the maximization of the spectral efficiency), is formulated as an integer (binary) linear optimization problem. Finally, a low- complexity clustering optimization algorithm for resource efficiency (CORE), is devised. Illustrative results prove that CORE, manages to increase the spectral efficiency and the network's capacity.

The efficient transmission of TCP traffic over OBS networks is a challenging problem, due to the high sensitivity of TCP congestion control mechanism to losses. In this paper, a traffic prediction scheme is proposed that exploits TCP traffic dynamics to optimize the performance of TCP transmission over OBS networks. Due to the TCP flow control mechanism, traffic dynamics can be accurately predicted in at least one RTT-long prediction window. In the proposed scheme, the prediction process is tightly coupled with the burst assembly process since the edge node is capable of inspecting incoming TCP traffic, keeping traffic statistics in parallel to the assembly process. These statistics are then used for traffic predictions. In this way burst size can be predicted and thus in advance reserve the appropriate resources. In this paper, we detail the traffic prediction mechanism and we also provide simulation results to assess its performance.

The efficient transmission of TCP traffic over OBS networks is a challenging problem, due to the high sensitivity of TCP congestion control mechanism to losses. In this work, we propose solutions to this problem, through the burst assembly and scheduling mechanism of OBS. We investigate and propose

In Optical Burst Switching (OBS) networks one of the most challenging problems is to provide service differentiation. The lack of buffering in the intermediate nodes, makes QoS provisioning in OBS networks a difficult problem to address. In this work, we propose a scheduling algorithm that supports

In this paper, we present a new hybrid switching architecture, termed as SLIP-IN, that combines electronic packet/burst with optical circuit switching. SLIP-IN architecture takes advantages of the pre-transmission idle periods of optical lightpaths and slips into them packets or bursts of packets. In optical circuit switching (wavelength-routing) networks, capacity is immediately hard-reserved upon the arrival of a setup message, but is only used after a round-trip time delay. This idle period is significant for optical multi-gigabit networks and can be used to transmit traffic of a lower class of service. In this paper, we present the main features and dependencies of the proposed hybrid switching architecture, and further we perform a detailed evaluation by conducting network wide simulation experiments on the NSFnet backbone topology. For this purpose, we have developed an extensive network simulator, where the basic features of the architecture were modeled. The extensive network study revealed that SLIP-IN architecture can achieve and sustain an adequate data rate with a finite worst case delay.