Multi-access Edge Computing (MEC) is a promising technology that provides low-latency processing capabilities. To optimize the network performance in an MEC system, an efficient routing path between a user and its serving MEC host is essential. The network performance is characterized by multiple attributes, including packet-loss probability, latency, and jitter. A user service may require a particular combination of such attributes, complicating the shortest-path computation. This paper introduces Q-Routing for Multi-Objective shortest-path computation (QR-MO), which simultaneously optimizes multiple attributes. We compare the QR-MO’s solutions with the optimal solutions provided by the Multi-objective Dijkstra Algorithm (MDA). The results show the favorable potential of QR-MO. After 100 episodes, QR-MO achieves 100% accuracy in networks with low to moderate average node degrees, regardless of the size, and more than 85% accuracy in networks with high average node degrees.

Multi-access Edge Computing (MEC) and network slicing technologies enable the Fifth Generation (5 G) of cellular networks to provide new revolutionary network services. In 5G-MEC systems, service providers are at the forefront of delivering cutting-edge, low-latency services that redefine connectivity experiences. Service providers face significant challenges in monetizing their offerings while ensuring the security and dependability of their network slices. The slice broker plays a crucial role in the supply chain, enabling efficient resource allocation and dynamic service provisioning to meet diverse user demands. There is a trade-off between revenue generation and resource allocation efficiency, which should be investigated to identify optimal revenue strategies for the slice broker. Security and dependability are paramount in 5G-MEC systems due to the extensive sharing of resources and critical data processing at the edge. From a security perspective, malicious attackers can be active to inject attacks, e.g., backdoor attacks, data poisoning etc. From a dependability point of view, there can be faults that may mislead the system, e.g., Byzantine faults. We propose an adaptive method by which agents can detect malicious attackers and faults during revenue model learning, where both security and dependability are considered.

The fifth generation of mobile networks can provide differentiated services enabled by network slicing and Multi-access Edge Computing. A new business actor, called Slice Broker (SB), is emerging as an intermediate entity that buys networking and computing resources from Infrastructure Providers (InPs) and provides network slices to Slice Tenants (STs). This paper addresses the problem of online jointly allocating all the network slices requested by the STs and selecting the resources to purchase from the InPs. The target of the problem is the minimization of the SB costs. We assume that a network slice is implemented as a sequence of Virtual Network Functions and that the InPs sell the networking and computing resources on predetermined configurations. The addressed problem is solved by using Deep Reinforcement Learning with a model-free policy gradient-based algorithm. The proposed solution is evaluated and compared with benchmark solutions in various scenarios. The results show that the benchmark solutions have a cost that is from 10% to more than twice higher than the proposed solution in all scenarios.

Multi-access Edge Computing (MEC) is an emerging technology that allows to reduce the service latency and traffic congestion and to enable cloud offloading and context awareness. MEC consists in deploying computing devices, called MEC Hosts (MEHs), close to the user. Given the mobility of the user, several problems rise. The first problem is to select a MEH to run the service requested by the user. Another problem is to select the path to steer the traffic from the user to the selected MEH. The paper jointly addresses these two problems. First, the paper proposes a procedure to create a graph that is able to capture both network-layer and application-layer performance. Then, the proposed graph is used to apply the Multi-objective Dijkstra Algorithm (MDA), a technique used for multi-objective optimization problems, in order to find solutions to the addressed problems by simultaneously considering different performance metrics and constraints. To evaluate the performance of MDA, the paper implements a testbed based on AdvantEDGE and Kubernetes to migrate a VideoLAN application between two MEHs. A controller has been realized to integrate MDA with the 5G-MEC system in the testbed. The results show that MDA is able to perform the migration with a limited impact on the network performance and user experience. The lack of migration would instead lead to a severe reduction of the user experience.

The fifth generation (5G) of mobile networks may offer a custom logical and virtualized network called network slicing. This virtualization opens a new opportunity to share infrastructure resources and encourage cooperation between several Infrastructure Providers (InPs) to offer tailored network slices for the Slice Tenants (STs). The Slice Broker (SB) is emerging as intermediate entity that purchases resources from the InPs and it offers network slices to the STs. The main challenge of the SB is to jointly decide the purchase of heterogeneous (data and network) resources from multiple InPs and create the slices to meet the various requests from the STs. Being an economical entity, the target of the SB is to maximize its profit by minimizing the costs while satisfying all the ST requests. This paper formulated the SB cost minimization problem and used CPLEX to obtain the optimal solution. The problem formulation considers the realistic scenario that the InPs offer the computing, storage and network resources by using predetermined configurations. Therefore, for each of the computing platform and logical connection, the SB may select one of the configurations. The proposed cost-minimization problem is compared with three alternative problems that have three different objectives: computing platform consolidation, network connection consolidation, and both computing-network consolidation. The computing platform and network connection consolidation are currently the most common approaches for decreasing resource costs. However, the result shows that consolidating computing and network resources fails to reach the actual minimal cost. The proposed problem finds the cheapest solution, which can save at least 30% of the total cost of the other approaches in every evaluated scenario. Moreover, consolidating the number of computing platforms can lead to the most expensive solution, up to 40% higher than the optimal solution of our proposed problem.

Multi-access Edge Computing (MEC) is one of the enabling technologies of the fifth generation (5G) of mobile networks. MEC enables services with strict latency requirements by bringing computing capabilities close to the users. As with any new technology, the dependability of MEC is one of the aspects that need to be carefully studied. In this paper, we propose a twolevel model to compute the availability of a 5G-MEC system. We then use the model to evaluate the availability of a 5G-MEC system under various configurations. The results show that having a single redundancy of the 5G-MEC elements leads to an acceptable availability. To reach a high availability, the software failure intensity of the management elements of 5G and MEC should be reduced.

Multi-access Edge Computing (MEC) enhances the capabilities of 5G by enabling the computation closer to the end-user for real-time and context-aware services. One of the main challenges of MEC is the migration of the MEC application in the presence of user mobility. MigraMEC is a hybrid testbed that simulates the network scenario and user mobility and emulates the MEC framework by using AdvantEDGE. Moreover, MigraMEC implements two physical MEC Hosts (MEHs) by using an extended version of Kubernetes (K8s). Finally, the MigraMEC controller interacts with both the emulative and experimental environments to ensure an efficient migration of the MEC application. Based on network information, the MigraMEC controller not only enforces the MEH where the MEC application is running but also the Point of Access (PoA) to which the user is connected. In our demonstration, the MEC application is a video streaming service, and the results highlight the need for multiple MEHs and efficient migration to maintain a high user experience.

Innovative services with strict requirements are expected in the fifth generation (5G) of mobile networks and beyond. For example, the Ultra-Reliable Low-Latency Communication (URLLC) requires up to 1 ms latency, end-to-end security, and reliability of up to 99.999%. The Multi-access Edge Computing (MEC) promises to support the delivery of URLLC services by providing computing and storage resources in the proximity of user equipment. The data which previously needed to be processed and stored in the cloud systems can be kept at the edge network, decreasing the total latency and increasing the context-awareness, security, and dependability. Vastly available resources, which are available from cloud to edge, must be appropriately allocated to deliver a service efficiently. The resource allocation problem in MEC for 5G-and-beyond networks can be formulated differently, depending on the nature of the problem. This survey outlines the resource allocation problem as a proper problem formulation, which can be addressed by target, resource type, resource issue, and the considered assumptions. Moreover, this paper also describes the open issues and future directions for MEC resource allocation based on the state of the art on this research topic.

The Network Slicing (NS) paradigm enables the partition of physical and virtual resources among multiple logical networks, possibly managed by different tenants. In such a scenario, network resources need to be dynamically allocated according to the slice requirements. In this paper, we attack the above problem by exploiting a Deep Reinforcement Learning approach. Our framework is based on a distributed architecture, where multiple agents cooperate towards a common goal. The agent training is carried out following the Advantage Actor Critic algorithm, which permits to handle continuous action spaces. By means of extensive simulations, we show that our approach yields better performance than both a static allocation of system resources and an efficient empirical strategy. At the same time, the proposed system ensures high adaptability to different scenarios without the need for additional training.

Network Function Virtualization (NFV) represents a key shift in nowadays network service provisioning by entailing higher flexibility, elasticity, and programmability of network services. Dependability is one of the main aspects that need to be investigated and tackled in order to profitably use NFV in the future. The main objective of this paper is to propose a comprehensive approach to estimate the end-to-end NFV-deployed service availability and present a quantitative assessment of the network factors that affect the availability of the service provided by an NFV architecture. To achieve this goal, we adopted a two-level availability model where i) the low level considers the network topology structure and NFV connectivity requirements through the definition of the system structure function based on minimal-cut sets and ii) the higher level examines dynamics and failure modes of network and NFV elements through stochastic activity networks. By using the proposed model, we have carried out an extensive sensitivity analysis to identify the impact on the service availability of the different service elements involved in the delivery, and their deployment across the network. The results highlight the significant impact that network nodes have on the end-to-end network service. Less robust network nodes may reduce the availability of an NFV-enabled service by more than one order of magnitude even though NFV elements like VNFs or MANO are provided with redundancy. Moreover, the results show that adopting an SDN-integrated network degrades the service availability and increases the vulnerability of the network service to SDN controllers unless adequately protected.

The fifth generation (5G) of cellular networks promises to be a major step in the evolution of wireless technology. 5G is planned to be used in a very broad set of application scenarios. These scenarios have strict heterogeneous requirements that will be accomplished by enhancements on the radio access network and a collection of innovative wireless technologies. Softwarization technologies, such as Software‐Defined Networking (SDN) and Network Function Virtualization (NFV), will play a key role in integrating these different technologies. Network slicing emerges as a cost‐efficient solution for the implementation of the diverse 5G requirements and verticals. The 5G radio access and core networks will be based on a SDN/NFV infrastructure, which will be able to orchestrate the resources and control the network in order to efficiently and flexibly and with scalability provide network services. In this paper, we present the up‐to‐date status of the software‐defined 5G radio access and core networks and a broad range of future research challenges on the orchestration and control aspects.

Software-defined networking (SDN) promises to improve the programmability and flexibility of networks, but also brings new challenges that need to be explored. The main objective of this paper is to include failure correlation in a quantitative assessment of the properties of SDN backbone networks to determine whether they can provide similar availability as the traditional IP backbone networks. To achieve this goal, this paper has formalized a two-level availability model that captures the global network connectivity without neglecting the essential details and which includes a failure correlation assessment. This paper proposes a modular and systematic approach for characterizing the principal minimal-cut sets in both SDN and traditional networks, and stochastic activity network models for characterizing the single network elements. To demonstrate the feasibility of the model, an extensive sensitivity analysis has been carried out on a national backbone network.

Software-Defined Networking (SDN) is a new paradigm that promises to enhance network flexibility and innovation. However, operators need to thoroughly assess its advantages and threats before they can implement it. Robustness and fault tolerance are among the main criteria to be considered in such assessment. The currently available SDN controllers offer different fault tolerance mechanisms, but there are still many open issues, especially regarding the trade-off between consistency and performance in a fault- tolerant SDN platform. In this paper, we describe existing fault-tolerant SDN controller solutions, and propose a mechanism to design a consistent and fault-tolerant Master-Slave SDN controller that is able to balance consistency and performance. The main objective of this paper is to bring the performance of an SDN Master-Slave controller as close as possible to the one offered by a single controller. This is obtained by introducing a simple replication scheme, combined with a consistency check and a correction mechanism, that influence the performance only during the few intervals when it is needed, instead of being active during the entire operation time.

The last 5 years have seen a dramatic shift in media distribution. For decades, TV and radio were solely provisioned using push-based broadcast technologies, forcing people to adhere to fixed schedules. The introduction of catch-up services, however, has now augmented such delivery with online pull-based alternatives. Typically, these allow users to fetch content for a limited period after initial broadcast, allowing users flexibility in accessing content. Whereas previous work has investigated both of these technologies, this paper explores and contrasts them, focusing on the network consequences of moving towards this multifaceted delivery model. Using traces from nearly 6 million users of BBC iPlayer, one of the largest catch-up TV services, we study this shift from push- to pull-based access. We propose a novel technique for unifying both push- and pull-based delivery: the Speculative Content Offloading and Recording Engine (SCORE). SCORE operates as a set-top box, which interacts with both broadcast push and online pull services. Whenever users wish to access media, it automatically switches between these distribution mechanisms in an attempt to optimize energy efficiency and network resource utilization. SCORE also can predict user viewing patterns, automatically recording certain shows from the broadcast interface. Evaluations using our BBC iPlayer traces show that, based on parameter settings, an oracle with complete knowledge of user consumption can save nearly 77% of the energy, and over 90% of the peak bandwidth, of pure IP streaming. Optimizing for energy consumption, SCORE can recover nearly half of both traffic and energy savings.

Network Function Virtualization (NFV) promises to transform the way telecom providers design and operate networks and network services. Virtualized Evolved Packet Core (vEPC) is one of the NFV use cases that has got most attention, where dependability is a major concern. In the traditional EPC, functions are deployed in proprietary network elements with proven characteristics, e.g., a defined availability, and corresponding guarantees. Hence, network operators have a firm basis for the design of a robust mobile core network. On the other hand, in the vEPC, network operators face a more challenging environment, where functions, subsystems and requirements are interrelated in a more complex manner. Hence, the assessment of the network robustness, and the design to meet dependability requirements become hard. In order to address this challenge, we provide initial guidelines to model system availability in vEPC scenarios, we propose a stochastic activity networks dependability model to assess it, and finally, we identify the most relevant factors to be considered by providers to fulfill the demanding dependability requirements of the mobile core.

"Catch-up", or on-demand access of previously broadcast TV content over the public Internet, constitutes a significant fraction of peak time network traffic. This paper analyses consumption patterns of nearly 6 million users of a nationwide deployment of a catch-up TV service, to understand the network support required. We find that catch-up has certain natural scaling properties compared to traditional TV: The on-demand nature spreads load over time, and users have much higher completion rates for content streams than previously reported. Users exhibit strong preferences for serialised content, and for specific genres.

The paper describes and compares different approaches that can be used to design Network Power Management methods, with the aim of reducing the power consumption of telecommunication networks. The approaches are based on the solution of optimization problems that have, in general, a Mixed Integer NonLinear Programming (MINLP) formulation. Given that the problems are NP-hard, exact methods for finding optimal solutions can be used only for scenarios of limited size. In this framework, the paper proposes a heuristic for finding a suboptimal solution of the Power Aware Routing and Network Design (PARND) problem, which is one of the more general design problems in Network Power Management. The simulation study highlights the capability of the proposed heuristic to obtain solutions near the optimum and to outperform the other approaches in terms of energy savings, while satisfying the constraints of the traffic demands.

Recently, the power consumption of the home/access networks has stimulated several research efforts devoted to the study of new mechanisms to optimize the energy utilization. In particular, most of these works is based on the idea that Ethernet links are mostly underutilized, and is thus aimed at realizing mechanisms able to adapt device consumption to the link utilization. Given the few works presenting an experimental study of the current Ethernet network devices from an energy point of view, this paper aims at presenting an extensive measurements campaign for evaluating the power consumption of two COTS Ethernet NICs. The measurements campaign, carried out following the procedures provided by the ECR Initiative™ and the SNE Energy Star, shows the energy inefficiency of the current NICs. Finally, using the experimental results, the paper presents a qualitative estimation of the energy savings achievable by implementing a well-known energy savings mechanism (i.e., ALR) in the Ethernet NICs.