The growing complexity of telecommunication networks increases the need for automated monitoring and detection of incidents. The increasing amount of data generated by current information and communication technologies equipment can be used to identify abnormal behavior and detect anomalies. In this paper, we propose a volume-based real-time solution for anomaly detection in operator networks. The proposal is based on a seasonal autoregressive integrated moving average (SARIMA) forecasting of time-series data, and the identification of outliers. We validate the proposed solution using real traffic traces from a nationwide operator’s network which we make publicly available. Our experimental results show that the proposed solution outperforms baseline approaches in terms of precision, being able to substantially reduce the amount of false positives, while maintaining the same level of recall.

Terahertz communications are envisioned as a key enabler for 6G networks. The abundant spectrum available in such ultra high frequencies has the potential to increase network capacity to huge data rates. However, they are extremely affected by blockages, to the point of disrupting ongoing communications. In this paper, we elaborate on the relevance of predicting visibility between users and access points (APs) to improve the performance of THz-based networks by minimizing blockages, that is, maximizing network availability, while at the same time keeping a low reconfiguration overhead. We propose a novel approach to address this problem, by combining a neural network (NN) for predicting future user–AP visibility probability, with a probability threshold for AP reselection to avoid unnecessary reconfigurations. Our experimental results demonstrate that current state-of-the-art handover mechanisms based on received signal strength are not adequate for THz communications, since they are ill-suited to handle hard blockages. Our proposed NN-based solution significantly outperforms them, demonstrating the interest of our strategy as a research line.

Physical-Layer Network Coding (PNC) is an effective technique to improve the throughput and latency in wireless networks. However, there are two major challenges for PNC, especially when using higher order modulations: 1) phase synchronization and power control at the paired User Equipments (UEs); and 2) the ambiguity removal of the PNC mapping at the relay node. To address these challenges, in this paper, we apply power control at transmitting UEs and exploit Reconfigurable Intelligent Surfaces (RISs) to synchronize the phase of the transmitted signals and ensure that they arrive at the relay with the same power and phase rotation. Then, we employ modular addition for an unambiguous PNC mapping for M-ary Quadrature Amplitude Modulations (M-QAM). We evaluate the performance of the system in the framework of Orthogonal Frequency Division Multiplexing (OFDM)-PNC for different RIS sizes and modulation orders. Furthermore, we study the sensitivity of PNC systems for Channel Estimation Error (CEE). The results reveal that 1) PNC systems show quite higher sensitivity to CEE compared with RIS-assisted one-way relay channel systems; 2) when the CEE is low, RIS can considerably enhance the Signal-to-Noise Ratio (SNR) of the PNC system, e.g., for a Bit Error Rate (BER) of 10−3(without channel coding), increasing the RIS size from one to 256 elements in 28 GHz band leads to 200% improvement in SNR.

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.

Abstract In this article, we evaluate the first experience of computation offloading from drones to real fifth‐generation (5G) operator systems, including commercial and private carrier‐grade 5G networks. A follow‐me drone service was implemented as a representative testbed of remote video analytics. In this application, an image of a person from a drone camera is processed at the edge, and image tracking displacements are translated into positioning commands that are sent back to the drone, so that the drone keeps the camera focused on the person at all times. The application is characterised to identify the processing and communication contributions to service delay. Then, we evaluate the latency of the application in a real non standalone 5G operator network, a standalone carrier‐grade 5G private network, and, to compare these results with previous research, a Wi‐Fi wireless local area network. We considered both multi‐access edge computing (MEC) and cloud offloading scenarios. Onboard computing was also evaluated to assess the trade‐offs with task offloading. The results determine the network configurations that are feasible for the follow‐me application use case depending on the mobility of the end user, and to what extent MEC is advantageous over a state‐of‐the‐art cloud service.

Multi-Access Edge Computing (MEC) is one of the fundamental technologies committed to satisfy the requirements targeted by 5G and beyond networks, such as low latency and massive communications. Nevertheless, deploying a large-scale MEC infrastructure will require a huge investment that should be minimized by optimizing the resources at edge locations, and to use centralized datacenters when possible. Thus, automated orchestration is essential for implementing mechanisms that deploy applications at the best location and even that relocate them, when necessary, to satisfy the Quality of Service (QoS) requirements. In this paper we describe an architecture for this purpose, which we have implemented in an experiment that demonstrates how Open Source MANO (OSM) can automate the relocation of a video processing application that helps drivers to remember the latest traffic sign viewed. Our proposal also includes two new components: the first one maintains the state of the applications when they are deployed at a new location, and the second one allows OSM managing the Open Network Edge Services Software (OpenNESS) edge platform. Finally, we elaborate on open challenges in MEC platforms.

Over the last few years the Multi-Access Edge Computing (MEC) paradigm has been gaining attention as a key enabler for low latency applications in cellular networks. In this paper we analyze a solution based on Software-Defined Networking (SDN) for supporting dynamic and transparent relocation of the endpoint of a communication from the core to an edge infrastructure in current cellular networks. We also provide results of real-world experiments utilising a Network Function Virtualization (NFV)-based testbed for evaluating session continuity and latency reduction when the gateway or anchor point used by two end User Equipments (UEs) is relocated to edge resources. Our experimental results show that the communication can be dynamically relocated from the core to the edge while guaranteeing session continuity during the whole process. We demonstrate that the mechanism is able to reduce latency considerably when the core network is congested.

Summary Network “softwarization” is enabling new features such as detecting the risk to violate a Service Level Agreement (SLA) and performing the required automated control actions on a network slice. A real‐time Network Function Virtualization (NFV) Management and Orchestration (MANO) system should detect any SLA degradation and act to recover the SLA. The closed‐loop automation approach monitors the state of the network and triggers policy‐based control actions when needed. The delay between the detection of an SLA issue and the completion of a correcting action is critical for satisfying latency‐related SLAs. Thus, in this article we propose this delay as a new key performance indicator (KPI) for NFV MANO systems, which we call Closed‐Loop Orchestration Delay (CLOD). CLOD is subdivided into independent time intervals that depend on the NFV MANO platform, the target Virtualized Infrastructure Management (VIM) and the target virtual network function (VNF) software. CLOD is evaluated on a real use case: a latency‐sensitive elastic network slice in which Internet user round‐trip time (RTT) is monitored. When the service level of the slice drops below a predefined limit a new data plane is deployed near the user reducing latency. In the evaluation, we considered two approaches: (i) a general implementation of the NFV MANO architecture based on the Open Network Automation Platform (ONAP) and OpenStack and (ii) an optimized ad‐hoc baseline implementation to minimize CLOD. We provide the details of both implementations and compare their CLODs leading to useful insights for the NFV MANO community.

Multi-Access Edge Computing (MEC) will allow implementing low-latency services that have been unfeasible so far. The European Telecommunications Standards Institute (ETSI) and the 3rd Generation Partnership Project (3GPP) are working towards the standardization of MEC in 5G networks and the corresponding solutions for routing user traffic to applications in local area networks. Nevertheless, there are neither practical implementations for dynamically relocating applications from the core to a MEC host nor from one MEC host to another ensuring service continuity. In this article we propose a solution based on Software-Defined Networking (SDN) to create a new instance of the IP anchor point to dynamically redirect User Equipment (UE) traffic to a new physical location (e.g., an edge infrastructure). We also present a novel approach that leverages SDN to replicate the previous context of the connection in the new instance of the IP anchor point, thus guaranteeing Session and Service Continuity (SSC), and compare it with alternative state replication strategies. This approach can be used to implement edge services in 4G or 5G networks.

Global optimization of access point (AP) assignment to user terminals requires efficient monitoring of user behavior, fast decision algorithms, efficient control signaling, and fast AP reassignment mechanisms. In this scenario, software defined networking (SDN) technology may be suitable for network monitoring, signaling, and control. We recently proposed embedding virtual switches in user terminals for direct management by an SDN controller, further contributing to SDN-oriented access network optimization. However, since users may restrict terminal-side traffic monitoring for privacy reasons (a common assumption by previous authors), we infer user traffic classes at the APs. On the other hand, since handovers will be more frequent in dense small-cell networks (e.g., mmWave-based 5G deployments will require dense network topologies with inter-site distances of ~150-200 m), the delay to take assignment decisions should be minimal. To this end, we propose taking fast decisions based exclusively on extremely simple network-side application flow-type predictions based on past user behavior. Using real data we show that a centralized allocation algorithm based on those predictions achieves network utilization levels that approximate those of optimal allocations. We also test a distributed version of this algorithm. Finally, we quantify the elapsed time since a user traffic event takes place until its terminal is assigned an AP, when needed.