The resolution of network intents to network service definitions is a key operation of intent-based networking. In this paper, we study the resolution of multi-domain network intents and propose EDAIR, an efficient distributed architecture that greatly improves accuracy of network intent resolution with little impact on performance. While previous state-of-the-art systems for multi-domain intent resolution consist on the sequential interrogation of all domains involved in the intent and the aggregation of their answers to construct a final network service definition, EDAIR defines an intelligent agent to represent each domain and constructs a distributed multi-agent system based on artificial intelligence. Only the required agents will be involved in intent resolution and the process stops as soon as a proper solution is achieved. We evaluated EDAIR to show that it is 48% more accurate and only increases the time needed to resolve intents by 38%, denoting its efficiency in the distributed operation.

Network virtualization offers enormous flexibility to construct network services tailored to the needs of network tenants and users. Network services can be implemented through composition of virtual network functions, which can be deployed in multiple sites, forming multi-site network services. However, network virtualization further induces the increasing complexity of the network, which, in turn, increases the difficulty of error avoidance in management operations. Thus, network management must be automated to tackle such complexity and eliminate errors. In this paper, we propose the intelligent network service automation system (INSA), and we demonstrate its high efficiency and accuracy in automated network service construction and management. INSA is designed to transparently orchestrate all elements of the management system using a control plane based on publish/subscribe messaging, known as a management service bus. This enables INSA to make highly accurate decisions using vast telemetry information without overloading the control plane. Additionally, INSA integrates key enablers such as efficient and scalable control plane networking, telemetry compression, distributed processing, hidden fault detection, and autonomic resource control. We evaluated INSA both theoretically and through a prototype implementation. The former proves our claims regarding the efficiency improvement provided by compression and distribution. The latter showcases INSA management accuracy and performance, and it also demonstrates how INSA manages multi-site network services. The evaluation results show that, in comparison with the state-of-the-art alternatives, INSA uses 15–36% less CPU time, possesses 80% higher scalability, and improves accuracy by 33%.

Network virtualization has become a central aspect of 6G networks, which enable the composition of virtual network functions (VNFs), which can be instantiated at single or multiple sites to create highly flexible virtual network services. As a result of the increased complexity imposed on network components, network service automation systems (NSASs) are expected to become key technology enablers for 6G, whose implementation requirements on the control plane that are not met by current approaches based on a central message broker. We propose to meet these requirements by extending NSAS with an efficient and scalable control plane network based on information-centric networking (ICN), named iCPN, which uses names instead of addresses for information transmission and exploits in-network cache. We evaluated an iCPN prototype, demonstrating that it uses 50% less bandwidth and 15-36% less computation time than common alternatives.

A network digital twin (NDT) is a virtual object that mirrors a network system (NS) so that a control and management system (CMS) may identify present and potential failures using artificial intelligence methods, resulting in enhanced NS resilience. In this study, we propose a method of enhancing an NDT by generating and transmitting telemetry knowledge objects (TKOs) instead of raw monitoring data. TKOs enforce data correction, provide rich information, and are inherently adaptable. We also apply the telemetry knowledge distributed processing (TKDP) approach to the CMS so that each NS element processes its own measurements to generate and transmit TKOs, thus avoiding the transmission of meaningless, wrongly formed, or incomplete data. Moreover, distributing the processing capacity allows the target NDTs to reflect more complex NS states than when using centralized processing. To maximize the impact of TKOs and TKDPs, we followed the current standardization specifications for network telemetry and NDT construction. Finally, we demonstrated the benefits of TKOs and TKDPs by evaluating the fidelity of an NDT that matches a typical scenario. We showed that the resultant NDT fidelity is much higher when using TKO and TKDP than when relying on current alternatives.

In this paper we propose a mechanism to automate the management of large networks based on the autonomic network management scheme. It incorporates a novel embedding algorithm that is 3 to 6 times faster than related work in finding out optimal embedding a virtual network (VN) or network slice (NS) to a substrate network (SN). It therefore finds out a configuration that assigns the elements of the VN/NS to the elements of the SN. The main novelty of this algorithm is the application of the knowledge from the VN/NS to get a configuration by improving a previous configuration, so less computing power is required than related work. The reasoning process is designed to consider the traffic of the network together with node load, so the resulting configuration optimizes the overall usage of the SN. Finally, the algorithm and heuristic also target the minimization of the number of operations required to obtain a new configuration from the current configuration, thus avoid the addition, removal, or movement of elements as much as possible. We demonstrate these claims by evaluating a proof-of-concept implementation of our solution and comparing it with the previous solutions.

The traditional telecommunications business is evolving towards offering a richer set of services beyond basic connectivity, leveraging on network programmability and virtualization. A versatile execution environment is required, capable of running different workloads in different locations in the network. Cloud computing is the key paradigm that allows fostering this trending change. One interesting question to solve is to what extent those computing environments have to move towards the network edge. Some services can be enabled by environments with increased capillarity, while others can be implemented in environments with more relaxed constraints (e.g., in terms of latency). This paper explores this topic by differentiating service edge from physical network edge and proposing an architecture based on the ALTO server for assisting orchestration systems in discriminating the suitable environments for each service. We present a network-flow strategy for assigning services to infrastructure elements following those precepts, together with an initial scalability evaluation of the proposed assignment solution to show its feasibility.

Network softwarization is a transformation trend for networks to converge in programmable service platforms able to host different requirements from multiple tenants. The concept of network slicing plays a key role in this trend. It offers isolated logical networks to different tenants over the same underlying infrastructure. Network slices hosting missioncritical services require fast reactions to changes in environment, adapting their capacity as needed to avoid service disruption and performance degradation. This represents a key challenge for the management plane, traditionally under the direct operation of human administrators, unable to provide a nimble enough response rate. To support tenants in driving lifecycle management operations over their network slices in a timely manner, datadriven, closed-loop mechanisms must be designed. In this paper we introduce a solution that takes advantage of events occurring outside the boundaries of a tenant-managed network slice to guide those adaptations. We finally demonstrate the feasibility of the proposal and describe an experimentation scenario to exploit such results.

The Autonomic Resource Control Architecture (ARCA) we have designed in previous work is intended to provide elastic resource adaptation in virtual computer and network systems by automating management tasks in edge/branch virtual networks. Our objective is to reduce the time required to continuously estimate or anticipate the amount of resources that a system requires by considering its workload together with event data notified by external detectors. However, analyzing such amount of data can take a long time and thus delay management decisions and system adaptation. Moreover, too frequent changes in the amount of allocated resources can hassle the underlying controllers. In this paper we propose a method to resolve these problems by filtering input data items to reduce their rate while smoothing non-persistent spikes to reduce the volatility of resource allocations. We evaluate it by running an algorithmic model with a set of input data following a Pareto distribution, analyzing the accuracy of different variations of the proposed method. We found that the accuracy of the resulting allocations is improved by 13% to 30%, depending on the specific configuration of the proposed method.

Vehicles are considered to constitute a large portion of new devices getting connected to 5G or beyond networks. Connected vehicles should be enabled to communicate with each other and roadside infrastructure securely, reliably, and in ultra-low latency. Vehicles should be able to discover and identify other vehicles in their surrounding very quickly. For this purpose, we propose a directory service that can store a huge number of vehicle's records and provide the records to querying vehicles in a very short latency (mostly less than 10 milliseconds). Similarly, it can dynamically update the records within a few seconds, while fully complying with record owner-centric privacy policy. The preliminary experiment results obtained from a simple testbed implementation with computers demonstrate the feasibility of the proposed system to meet the target lookup and update performance of connected vehicles.

Current network service requirements are increasing the pressure on the level of flexibility and reliability that must be provided by the virtual computer and network systems that host them while enforcing close-to-optimum resource allocations to minimize both monetary and operational costs. A key requirement is the fast detection and adaptation to workload changes, which is beyond human abilities, so the operation must be automated. To do so we propose to use an autonomic resource control architecture. In this paper, we formalize the architecture and discuss how it is able to detect the CPU load changes of a networked service by analyzing heterogeneous observations, including both resource load measurements and external events, and adapt (increase or decrease) the allocated resources consequently. The key benefit of our proposal is that, in most situations, including external information into the analysis, such as seismometer measurements, allows the control architecture to detect new situations earlier than analyzing only load measurements. We demonstrate such claims with the results obtained from the execution of a proof-of-concept implementation of the architecture we propose on an emergency scenario.

In this paper, we propose a mechanism for autoconfiguration of name-bound virtual networks (NBVNs) for Internet of Things (IoT). Some IoT standardization groups have defined APIs for IoT device communications, in which they indicate the correspondent nodes and resources by names. However, current technologies for the construction of Virtual Networks (VNs) rely on VLANs, IP routing, and OpenFlow control, thus they do not provide a name-based solution. Our proposal fills this gap. We first define business players that construct and use the NBVNs. They are Application Service Provider (ASP), Virtual Network Operator (VNO), and Infrastructure Provider (InP). We then specify the roles of ASP/VNO/InP, including the tasks that must be performed by their managers, and the required interactions for the purpose of autoconfiguration. Subsequently, we describe the system we developed to implement the operations of ASP/VNO/InP and thus construct NBVNs automatically. For each NBVN, the required IPv6 addresses are automatically allocated and assigned to the network nodes, including the IoT devices. Moreover, data forwarding and name resolution mechanisms are also configured automatically. Thus, the proposed system constructs area-and/or time-bound VNs for offering network services to event-centric IoT applications, such as outdoor concerts and sporting events. Finally, we demonstrate that the constructed NBVNs are capable to configure thousands of addresses and name entries within a minute, thus IoT devices can communicate among them by using their correspondent node names instead of their addresses. Moreover, both ASP and VNO can operate the servers, switches, and routers pertaining to the NBVNs by also using their names.

The raise of Software Defined Networking (SDN) opens a wide spectrum of improvements for network architectures targeting different problems. It includes the case of those architectures designed to overcome mobility and multi-homing problems in big networks, such as the Internet, as is the case of the HIMALIS network architecture. In this paper we present an approach for HIMALIS to benefit from SDN technologies by changing the way it addresses its control operations. We propose to leverage control operations of the HIMALIS architecture through an identity-based control plane that has been designed following the principles of SDN. Thus, our approach proposes to integrate the main elements of HIMALIS with a controller, which provides enormous benefits to the architecture in general, such as the possibility to perform control operations with global knowledge of the network status and the evolution of the functions provided by the control plane without changing the implementation of the end nodes. Finally, in order to demonstrate those benefits, we have built architectonic models for both HIMALIS and our proposal, executed different simulations with them running on different mobility scenarios, and extracted the results that show clear improvement of our proposal.

The GEMBus framework is intended to provide the necessary mechanisms to build federated composite services. Thus, it can constitute a key element in achieving cross-stratum optimization by combining network-layer events with service-layer operation knowledge. Taking advantage of the unique service perspective of the high level network operations we can use the GEMBus framework to compose the services with the network elements in order to build an enhanced and network-aware service. In this paper we give a brief description of the GEMBus architecture and discuss how to use it to build such services. We also evaluate the architecture using a proof-of-concept implementation and discuss the obtained results.

In this paper we show a service-oriented architecture for managing network systems and services without human intervention, ensuring compliance with rules and policies set by administrators. Furthermore, we show the evaluation results of a proof-of-concept implementation of the architecture running in a test environment where it has to manage the links that connects the main router of an organization to the backbone of its provider, following simple policies that fixes saturation situations and improve network availability.

We investigate the use of adaptive multimedia applications to improve the user-perceived QoS in next-generation wireless networks. These networks will be characterized by the heterogeneity of the access technologies including ad hoc wireless network extensions. In these network scenarios in which the links are error-prone, their capacity is extremely variable and even the topology can change due to handovers and mobility of the nodes, it is unrealistic to think of traditional network-layer QoS mechanisms alone, as a means to offer strict end-to-end QoS guarantees. As a complement to the network-layer QoS protocols, this paper presents an intelligent adaptation middleware allowing multimedia applications to dynamically adapt their internal settings. This intelligence enables the applications to minimize the impact of the adverse and changing network conditions in the QoS level perceived by the user. We use a machine learning technique to model the user's perception of QoS from a large set of QoS scores given by real users. We present some results demonstrating that our proposal clearly outperforms traditional multimedia internetworking.