• Large-scale 5G mobility dataset generation via a comprehensive simulation workflow. • Predictive handover: Development of an LSTM model to forecast future serving cells. • Proactive tracking: High accuracy trajectories in variable traffic density scenarios. • Smart migration: proactive decisions for low latency and high resource efficiency. In this work, we address the challenge of maintaining high Quality of Service (QoS) of edge services provided to constantly moving devices, especially in urban environments such as connected vehicles. To support proactive service migration and maintain low latency in these environments we propose a long short-term memory model that predicts the end device trajectory in terms of future base stations. In our work, we have used traffic data from urban environments combined with wireless channel metrics from cellular networks to be able to generate an efficient predictive model that works well with varying traffic densities. Our experimental results show that the proposed model offers high accuracy in predicting an array of future coverage cells while remaining resilient to constantly changing traffic volumes and patterns. This study highlights the importance of unifying artificial intelligence and machine learning for dynamic resource orchestration and ensuring QoS in real-time applications operating on smart 6G networks.

6G networks aim to offer users ultra-efficient access to networked services based on seamless integration of the network with low latency solutions such as the Multi-access Edge Computing. To achieve this goal, the 6G edge orchestration should support smart decisions based on the learned user mobility behaviour. In this paper, we focus on the handover prediction problem that is an essential component to implement proactive service migration in 6G networks, particularly when migration delays require reliable forecasts several seconds ahead. We analyse three novel models based on Long Short-Term Memory (LSTM) neural networks aiming to improve the handover prediction performances when working with a wider prediction horizon. The architectures include: a parallel multi-output model that produces all outputs simultaneously, a multi-output model with the addition of a Bahdanau attention mechanism that works by iterative prediction of future time points, and an ensemble of single-output models that incorporates previous predictions into future steps. Our results show that the single-output ensemble models provide stable results over the entire prediction horizon, although prediction time costs must be assumed. The iterative attention model achieves a balance in mid-range predictions, while the parallel multi-output model stands out for its excels in early-horizon prediction. These findings highlight a clear trade-off between predictive accuracy and computational cost, indicating that the MEC orchestrator should choose an architecture guided by the requirements of the targeted MEC service and the lead time available for proactive migration. • 6G MEC handover prediction with Bi-LSTM and extended horizon in dynamic urban areas. • Comparative analysis of handover prediction with multi and single output attention. • Balancing accuracy and resources between multi and single-output models. • Long-term handover prediction strategy recommending optimal modeling approaches based on prediction horizon length.

Abstract The recent introduction of full‐mesh virtual private network (VPN) solutions which offer near native performance, coupled with modern encryption algorithms and easy scalability as a result of a central control plane have a strong potential to enable the implementation of a seamless edge‐cloud continuum. To test the performance of existing solutions in this domain, we present a framework consisted of both essential and optional features that full‐mesh VPN solutions need to support before they can be used for interconnecting geographically dispersed compute nodes. We then apply this framework on existing offerings and select three VPN solutions for further tests: Headscale, Netbird, and ZeroTier. We evaluate their features in the context of establishing an underlay network on top of which a Kubernetes overlay network can be created. We test pod‐to‐pod TCP and UDP throughput as well as Kubernetes application programming interface (API) response times, in multiple scenarios, accounting for adverse network conditions such as packet loss or packet delay. Based on the obtained measurement results and through analysis of the underlying strengths and weaknesses of the individual implementations, we draw conclusions on the preferred VPN solution depending on the use‐case at hand, striking a balance between usability and performance.

When moving cloud computing towards the edge in order to provide low latency and location-based Internet services, there are additional considerations that need to be introduced in the traditional cloud approaches to adapt them to fog computing. This paper discusses a proposal for resource allocation and migration strategies adapted specifically to fog computing. The proposal is based on the concept of virtual and physical communities and it has already shown its efficiency when applied to cloud computing. Simulation-based scenarios provide the performance of our community-based proposal following the geographical trajectory of mobile users.

The need for customer-centric agile integrated environment offered in an on-demand self-service manner challenges the traditional R&E operations support systems. The service provider architecture proposed in this paper aims to satisfy end-user's requirements by providing a coordinated provisioning of virtualized resources through automation. Empowered by the establishment of open collaborative partnerships it offers expansion of the service catalog with multi-domain solutions based on standardized interoperability between R&E and commercial CSPs. The benefits of the approach are presented via a cloud based case study service that is transformed from a traditional silo to a business process modeled service with added values.

Nowadays, companies collect data at an increasingly high rate to the extent that traditional implementation of algorithms cannot cope with it in reasonable time.On the other hand, analysis of the available data is a key to the business success.In a Big Data setting tasks like feature selection, finding discretization thresholds of continuous data, building decision threes, etc are especially difficult.In this paper we discuss how a parallel implementation of the algorithm for computing the information gain can address these issues.Our approach is based on writing Pig Latin scripts that are compiled into MapReduce jobs which then can be executed on Hadoop clusters.In order to implement the algorithm first we define a framework for developing arbitrary algorithms and then we apply it for the task at hand.With intent to analyze the impact of the parallelization, we have processed the FedCSIS AAIA'14 dataset with the proposed implementation of the information gain.During the experiments we evaluate the speedup of the parallelization compared to a one-node cluster.We also analyze how to optimally determine the number of map and reduce tasks for a given cluster.To demonstrate the portability of the implementation we present results using an on-premises and Amazon AWS clusters.Finally, we illustrate the scalability of the implementation by evaluating it on a replicated version of the same dataset which is 80 times larger than the original.

This paper presents results of the ongoing development of the Open Cloud eXchange (OCX) that has been proposed in the framework of the GN3plus project. Its aim is to provide cloud aware network infrastructure to power and support modern data intensive research at European universities and research organisations. The paper describes the OCX concept, architecture, design and implementation options. OCX includes 3 major components: distributed L0-L2 (optionally L3) network infrastructure that includes OCX points of presence (OCXP) interconnected with GEANT backbone; the Trusted Third Party (TTP) for building dynamic trust federations; and the marketplace to enable publishing and discovery of cloud services. OCX intends to be neutral to actual cloud services provisioning and limits its services to Layer 0 through Layer 2 in order to remain transparent to current cloud services model. The recent developments include an architectural update, API definition, integration with higher-level applications and workflow control, signaling and intercloud topology modelling and visualization. The paper reports about results and experiences learnt from the recent OCX demonstrations at the SC14 Exhibition in November 2014 that demonstrated the benefits of an OCX enabled Intercloud infrastructure for running data intensive real-time cloud applications on top of the advanced GEANT multi-gigabit network. The implemented OCX functionality allowed applications to control the network path for data transfer and service delivery connectivity between multiple Cloud Service Providers (CSPs). It was used in combination with a multi-cloud workflow management and planning application (Vampire) that enables data processing performance monitoring and migration of VMs and processes to an alternative location based on performance predictions.

This paper presents the concept of Open Cloud eXchange (OCX) that has been proposed to bridge the gap between two major components of the cloud services provisioning infrastructure: Cloud Service Provider (CSP) infrastructure, and cloud services delivery infrastructure which in many cases requires dedicated local infrastructure and quality of services that cannot be delivered by the public Internet infrastructure. In both cases there is a need for interconnecting the CSP infrastructure and local access network infrastructure, in particular, to solve the "last mile" problem in delivering cloud services to customer locations and individual (end-)users. The OCX remains neutral to actual cloud services provisioning and limit its services to Layer 0 through Layer 2 to remain transparent to current cloud services model. The proposed document identifies the initial set of requirements to OCX, that can be run by NRENs, as a part of the G&201;ANT network, or jointly, and provides suggestions about OCX implementation. The proposed OCX concept will leverage the existing Internet eXchange (IX) and GLIF Open Light path Exchange (GOLE) solutions and practices, adding specific functionality that will simplify inter-CSP and customer infrastructure integration when supporting basic cloud services provisioning models, in particular Trusted Third Party (TTP) services to allow federated infrastructure and access control, commonly used by NRENs. The paper also describes trusted/secured topology exchange protocol and dynamic trust establishment protocol as a part of the OCX services.