The transition from traditional High Performance Computing (HPC) to the Computing Continuum emphasizes efficient resource management and sustainable practices across Multi-Scale hybrid architectures. This paper introduces a multidimensional metric framework to characterize these systems and guide deployment strategies for modern workloads. The framework combines Architectural Performance metrics (such as Throughput, Latency, Scalability), System Utilization, and key Sustainability and Accuracy indicators (such as Energy Efficiency and Power Consumption). Using a modular hybrid testbed, experiments reveal complex relationships among metrics, especially the trade-offs between accuracy and energy, and the efficiency of hybrid nodes. The guidelines help identify optimal operating points and lay the groundwork for improving orchestrators and schedulers (e.g., Kubernetes) to assign demanding applications, including AI and Quantum Computing, to suitable system modules, ensuring high performance and sustainability.

We discuss a Quantum-Enhanced Computing Continuum, a heterogeneous, hybrid architecture that integrates quantum processing units (QPUs) within an Edge-Cloud-HPC fabric. Promote sustainability by shifting from performance to "energy-aware integration.' The architecture has three layers: a Physical Layer with shared fiber-optic infrastructure, a Control and Orchestration Layer managed by the user, and an Application Layer with an Adaptive Quantum Classical Fusion (AQCF) framework. Tighter system integration, like moving from cloud coupling to cryogenic logic, reduces energy waste and "thermal footprints.' The aim is a Green Performance Advantage: energy per problem solved in the era of Advanced Computing.

Edge, Fog, and Cloud-HPC infrastructures require innovative solutions to address energy constraints, workload variability, and scaling requirements. This paper presents adaptable architectures using milliclusters-lightweight compute groups that can be dynamically created, combined, or dissolved to meet application and energy demands while supporting sustainability on the Continuum. The architecture blends millicluster flexibility with communication-aware orchestration, using multitier interconnects and topology-sensitive scheduling to reduce data transfer and improve efficiency. Early results indicate that adaptive millicluster creation and communication-focused execution enhance performance stability and energy efficiency across workloads, underscoring the importance of malleability for the future of the Computing Continuum.

This article presents Named Resource Networking (NRN), a resource reservation system that allows user to formulate resources interest. Every available resources (computing, storage and network) is given a unique name in a NDN manner. NRN is an end-to-end reservation system that provides a way to discover resources through the network and formulates some network requirement to reach the reserved resource. It leverages from the computing continuum context being able to exploit resources anywhere on the network, being edge equipment, networking equipment from the infrastructure or cloud.

Computing Continuum, with its diverse architectures and distributed features, presents certain challenges in managing and orchestrating resources. Traditional orchestration often requires specialized knowledge, which can be a barrier for users who may lack expertise. In this paper, we introduce User-Friendly Orchestration Management (UFOM), an innovative approach designed to simplify the orchestration of containerized services. With an intuitive interface, automated workflows, and contextual support, UFOM streamlines the entire process. Importantly, UFOM integrates seamlessly with Osmotic Computing, enabling smooth interactions between users and technology. This novel approach democratizes orchestration, empowering users to deploy and manage applications with minimal technical difficulty. We will examine its impact on Quality of Experience (QoE) and present a case study on smart home automation. This case study highlights the capabilities of UFOM in real-world environments, stressing its effectiveness in streamlining deployment, monitoring, and maintenance. Our findings suggest that UFOM not only fosters user independence, but also maintains system reliability and efficiency.

This preliminary study describes the renewable energy production of the Fenicia district in the city of Bogota, Colombia. Based on urban distribution and population, it focuses on electrical production through solar panels and biogas production from waste recycling through methanation and its conversion into electricity. The objective of studying this energy mix is to size and design a micro datacenter for the district, following the principles of a circular economy and so powered exclusively by locally produced renewable energy. The study reveals an electricity production preference towards solar panels by an order of magnitude greater than methanation, although the interest in methanation remains significant due to its much lower fluctuations.

The integration of Artificial Intelligence (AI) and Ambient Intelligence (AmI) has emerged as a promising approach to creating responsive and contextually aware environments. AmI creates contextually aware environments by seamlessly integrating intelligent technologies, while AI develops algorithms for autonomous learning and decision-making. However, embedding AI within AmI environments faces challenges due to limited resources and energy constraints. While recent research on embedded AI has primarily focused on specific tasks of AmI, our goal is to develop a comprehensive framework encompassing all the necessary components for practical use cases. Through this endeavor, we aim to explore power-aware designs and distributed learning as fundamental approaches to address limited computational resources, energy constraints, and dynamic context variations challenges.

The evolution of the Computing Continuum, coupled with DevOps practices, marks a significant transformation in modern computing. This paper examines the integration of cloud, fog, edge, and IoT technologies to enhance resource utilization, scalability, and collaboration. The synergy between DevOps and orchestration systems automates essential processes, optimizing both performance and security. Despite challenges such as coordination complexities and talent shortages, these advancements hold the potential for increased flexibility and efficiency in Computing Continuum environments. The paper concludes by proposing a definition of the Computing Continuum, informed by state-of-the-art concepts and the interplay between multi-architecture orchestration and DevOps culture.

High-Performance Computing – HPC platforms based in post-Moore architectures integrate multiple specific-purpose chips in heterogenous architectures, focusing on embedded and low-power computational infrastructures – i.e., RISC architecture processors – to improve energy efficiency and low cost. This characteristic allows us to consider a certain search for computing sustainability. Currently, these platforms are very useful to implement from miniclusters to edge/cloud exchange nodes because they present an adequate relationship in terms of performance, scalability, and sustainability in addition to their low cost. This work discusses the performance and sustainability of deploying system operative images and applications in embedded post-Moore architectures addressed to HPC sustainable platforms.

In recent years, emerging computing paradigms have paved the way for the development of the Computing Continuum, a concept that enables applications to efficiently allocate geo-distributed resources across the network. Despite its potential, fully realizing the Computing Continuum remains a challenge due to the lack of suitable research infrastructures that support experimentation at scale. To address this gap, we propose a conceptual testbed designed to enhance the replicability, scalability, and robustness of Computing Continuum and network operating system experiments. Our testbed provides a high degree of flexibility by allowing experimenters to modify the operating systems of network equipment and dynamically reconfigure network topologies. To illustrate its versatility, we define three distinct usage scenarios, ranging from multi-operator environments to internal network architectures within telecommunication providers, all of which can be deployed on the proposed network topology. Additionally, this article explores solutions for virtual topology management, operating system deployment, and service orchestration.

Clustering is an effective method for detecting network intrusions, especially due to its unsupervised nature and fast reactivity, which is particularly advantageous for the identification of zero-day attacks. Despite its potential, Intrusion Detection Systems (IDS) often uses clustering algorithms based on their simplicity, without rigorous comparative analysis, and ignores important limitations of some algorithms such as preset cluster numbers and inflexible cluster shapes. These shortcomings hamper the effective use of network traffic data to detect intrusion, especially given its dynamic nature. In this article, we evaluate the effectiveness of various clustering algorithms—specifically partitioning, hierarchical, and density-based methods—in detecting anomalies in network traffic. We focus on analyzing the formation and characteristics of anomaly clusters within these approaches and propose a complete evaluation methodology suitable for intrusion detection using streaming data to be closer to real-world scenarios. Our results reveal that no single clustering algorithm universally excels on every dataset, prompting us to introduce an ensemble clustering approach that brings together the strengths of different classes of algorithms. Preliminary tests demonstrate the superior performance of our approach on three IDS datasets with a good accuracy and F1-score greater than 0.8 for all of them.

Intrusion detection systems (IDS) based on deep learning have proven successful, but struggle to learn continuously and detect new attacks over time due to a supervised label-based reward function. In this article, we introduce an unsupervised Deep Double Q Learning (DDQL) method that aims to detect attacks and learn new behaviors through an unsupervised reward function leveraging a normality score inspired by car traffic anomaly detection.

This research empirically examines embedded development tools viable for on-device TinyML implementation. The research evaluates various development tools with various abstraction levels on resource-constrained IoT devices, from basic hardware manipulation to deployment of minimalistic ML training. The analysis encompasses memory usage, energy consumption, and performance metrics during model training and inference and usability of the different solutions. Arduino Framework offers ease of implementation but with increased energy consumption compared to the native option, while RIOT OS exhibits efficient energy consumption despite higher memory utilization with equivalent ease of use. The absence of certain critical functionalities like DVFS directly integrated into the OS highlights limitations for fine hardware control.

Crude oil leakages and spills (OLS) are some of the problems attributed to pipeline failures in the oil and gas industry’s midstream sector. Consequently, they are monitored via several leakage detection and localisation techniques (LDTs) comprising classical methods and, recently, Internet of Things (IoT)-based systems via wireless sensor networks (WSNs). Although the latter techniques are proven to be more efficient, they are susceptible to other types of failures such as high false alarms or single point of failure (SPOF) due to their centralised implementations. Therefore, in this work, we present a hybrid distributed leakage detection and localisation technique (HyDiLLEch), which combines multiple classical LDTs. The technique is implemented in two versions, a single-hop and a double-hop version. The evaluation of the results is based on the resilience to SPOFs, the accuracy of detection and localisation, and communication efficiency. The results obtained from the placement strategy and the distributed spatial data correlation include increased sensitivity to leakage detection and localisation and the elimination of the SPOF related to the centralised LDTs by increasing the number of node-detecting and localising (NDL) leakages to four and six in the single-hop and double-hop versions, respectively. In addition, the accuracy of leakages is improved from 0 to 32 m in nodes that were physically close to the leakage points while keeping the communication overhead minimal.

Virtual threads are a highly anticipated feature in the Java world, aiming at improving resource efficiency in the JVM for I/O intensive operations while simplifying developer experience. This feature keeps the traditional thread abstraction and makes it compatible with most of the existing Java applications, allowing developers preferring synchronous imperative abstractions to benefit from better performance without switching to asynchronous and reactive programming models. However, limitations currently hinder the usability of virtual threads. These limitations must be considered when building a piece of software around virtual threads for they might have non-trivial effects. This paper (i) discusses the different strategies envisioned to leverage virtual threads in the Quarkus framework, (ii) gives an overview of the final implementation, (iii) presents the benchmark used to characterize the benefits of using virtual threads in a typical container environment where resources are scarce compared to using Quarkus with traditional thread pools and Quarkus with reactive libraries ; (iv) results are interpreted and discussed. Our study reveals that the integration of virtual threads in Quarkus doesn't perform as well as Quarkus-reactive. This seems to be due to a mismatch between the core hypothesis of Netty and virtual threads regarding the amount of threads available.

The training energy efficiency of deep neural networks became an extensively studied research topic in the last years. Some of the existing approaches seek to reduce the size of the architecture by either starting the training with a large network and pruning it, or by beginning with a seed architecture and then growing it. Instead of compressing the architecture, other approaches aim to reduce the number of training examples through data selection. While various approaches belonging to these two categories have been proposed, only a few works actually conduct energy measurements. Others merely mention potential gains in efficiency or rely on alternative evaluation metrics such as FLOPs. In this paper, we conduct a series of experiments both on a synthetic dataset and on image classification benchmarks in order to compare the impact of pruning, architecture growing and data selection on training energy consumption and prediction quality. Our results show that growing maintains a high prediction quality but brings limited energy gains when the size of the resulting architecture is large. Pruning can offer high gains, but also impacts accuracy, making it more suited for large models. Data selection provides energy gains correlated with the selectivity rate but causes an accuracy loss. We find that the effectiveness of every technique depends on its hyperparameters and on the architecture size.

The proliferation of Internet of Things (IoT) devices has intensified the demand for energyefficient solutions supporting ondevice and distributed learning applications. This re search presents a circumscribed comparative analysis of radioenabled ultralow power IoT devices, specifically focusing on their suitability for computationheavy use cases. Our analysis centers on middleend IoT devices that serve as a vital interface between the Electronics and Machine Learn ing communities. The evaluation encompasses a diverse range of IoT hardware equipped with inte grated radios. We established functional datasheetbased criteria completed with accessibility and communitywise criteria to study and offer valuable insights into each selected node's performance tradeoffs, strengths, and weaknesses. This study provides crucial guidance for TinyML practitioners seeking to make informed device selections for their applications.