Towards Autonomous Buildings, Communities and Positive Energy Districts: Multi-Layer Modeling and Edge-Enabled Islanding for the Energy Transition
Publication: IEECB&SC’26
Authors: Brian O’Regan, Farah Tahir, Karen Mould, Eoin O’Leidhin
Read Abstract
Modern electricity grids demand architectures that are not only efficient but also resilient, adaptive, and intelligent as the energy system transitions toward decarbonisation, decentralisation, and digitalisation. As the share of renewable energy sources grows, and as extreme weather events become more frequent and harder to predict, and unpredictable, the conventional one-way relationship between buildings, communities, and the grid is being replaced by a dynamic ecosystem of energy prosumers, flexumers, local flexibility markets, and Positive Energy Districts (PEDs). New integrated approaches are required to model, coordinate, and control these systems while maintaining stability under uncertainty.
This paper presents an integrated multi-layer modelling framework that unites deterministic, stochastic, and artificial intelligence (AI) approaches within a common architecture designed for electricity grid multi-islanding operation. This allows for different parts of the grid to be isolated from the main grid during brown or blackouts, reducing downtime for users and increasing reliability and confidence throughout the system. The deterministic layer focuses primarily on predictability and reproducibility, and enforces physical and operational feasibility through thermodynamic, electrical, and control constraints. Whereas the stochastic layer captures the inherent uncertainty from renewable variability, market volatility, and occupant behaviour using probabilistic and ensemble models. The AI layer enhances predictive accuracy and computational efficiency through data-driven forecasting and surrogate modelling. A reinforcement learning layer based on Q-learning introduces adaptivity and self-correction, enabling autonomous optimisation of control policies and real-time response to evolving conditions.
Ultimately, these layers work together in a structured, hierarchical way. The lower-level deterministic constraints set the physical and operational limits of the system, defining what solutions are feasible. Stochastic models capture and quantify uncertainties, such as fluctuating renewable generation or the changing demand. Artificial Intelligence (AI) enhances predictive accuracy by processing large datasets quickly, while the top-level Q-learning continuously refines control strategies to optimise system performance within the given limits.
Together, these layers enable resilient, federated energy clusters (energy communities or PEDs) that can operate independently or cooperatively with the grid. Through fragmented islanding, communities dynamically segment into autonomous nodes during disturbances, maintain local power balance, and later resynchronise to the main grid. The framework is demonstrated through case studies in Ireland and Croatia, validating applicability across diverse European energy contexts. This multi-layer, edge-IoT enabled paradigm positions buildings and communities as active participants in an intelligent grid capable of learning, adapting, and trading energy in real time via peer-to-peer (P2P) markets.
Relevance of the Paper to the Project:
This paper presents the work being carried out by UCC in O-CEI Pilot 1.1 (Ireland). It covers the symbiotic buildings and the applications being developed, including: FLEXUS, PARA//EL, EdgeWare, and minutemanSEM. It also goes into the multi-layered modelling, Response Derivatives (RD), fragmented islanding, and FlexusHive. It also has information on Pilot 1.1 and Pilot 1.2.
The Role of Continual Learning in the Cloud-Edge Continuum: A Review on Efficiency and Trustworthiness
Publication: Integrative Journal of Conference Proceedings (Int J Conf Proc), Volume 4, Issue 3, January 09, 2026
Authors: Andrés L. Suárez-Cetrulo, Rajnish Rakholia, Miguel Aspis, Jaydeep Samanta, Cristian Bosch and Ricardo Simón Carbajo
Read Abstract
The Cloud-Edge Continuum (CEC) represents a paradigm shift towards a heterogeneous, distributed computing landscape. This environment is characterized by massively distributed data sources, dynamic network conditions, and fluctuating computational loads. Traditional Machine Learning (ML) models, trained offline in a centralized manner, are not suited for this reality. They fail to adapt to the constant stream of new data, making them vulnerable to concept drift. This leads to inevitable performance degradation, creates significant processing bottlenecks, and undermines core Trustworthy AI principles of robustness and reliability. This paper argues that Continual Learning (CL) is a critical and necessary paradigm for robust and efficient intelligence in the CEC. We review the relevance of CL, data stream learning, and integrated concept drift detection as the primary mechanisms for maintaining model robustness and resilience. CL, implemented through a combination of data-centric and model-centric compression and frugal AI techniques, is vital for achieving both the efficiency and trustworthiness demanded by next-generation applications operating in the CEC. These methodologies include iterative fine-tuning, model compression, knowledge distillation, and dynamic neural network growth. This adaptive intelligence is required not only for end-user applications but also for MetaOS-level orchestration across the Cloud-Edge Continuum. This paper concludes by presenting key findings that highlight the essential role of adaptive learning across the continuum and outlines future research directions aimed at enabling scalable, trustworthy, and resource-efficient continual learning for MetaOS-based orchestration and management.
Relevance of the Paper to the Project:
The paper analyzes intelligence and orchestration across several European-funded MetaOS projects, including ICOS, NEMO, aerOS, NEPHELE, NebulOuS, and FLUIDOS, which are incorporated into the broader O-CEI ecosystem. The paper also addresses critical AI elements required for dynamic environments, such as continual learning, concept drift detection, parameter-efficient fine-tuning, and frugal AI models which can directly be leveraged as utility-based services to ensure that AI models remain robust, reliable, and energy-efficient at the network edge in O-CEI.
“Securing Data-Driven Cognitive V2G Charging: Edge Intelligence and Cybersecurity for Trusted EV Energy Exchange”
Authors: Maria Makrynioti, George Lazaridis, Georgios Spanos, Georgios Stavropoulos, Periklis Chatzimisios, Silvia Canale, Esther Stallone, Konstantinos Votis, Dimitrios Tzovaras
Read Abstract
The rapid evolution and usage of Electric Vehicles (EVs) and bidirectional Vehicle-to-Grid (V2G) technologies is reshaping the energy ecosystem, creating new opportunities for data-driven optimization while exposing charging infrastructures to evolving cybersecurity risks. This paper presents a conceptual framework for Securing Data-Driven Cognitive V2G Charging that leverages edge intelligence, distributed machine learning, and 5G-enabled IoT microservices to enable trusted EV energy exchange. Building on prior knowledge through European Union (EU) funded projects, the proposed approach addresses two complementary scenarios: (i) cognitive edge optimization of power flows for intelligent and cost-efficient EV charging under volatile renewable generation, and (ii) cybersecurity and trust enhancement in V2G data exchange through continuous monitoring, vulnerability detection, and secure, auditable data workflows. By integrating cognitive decision-making with big data analytics, the framework enables measurable cost savings across the EV charging value chain, while simultaneously ensuring grid stability, power quality, and resilience against cyber threats. Furthermore, the paper discusses open challenges and research directions for building secure, scalable, and trustworthy V2G charging infrastructures, highlighting the role of big data–driven cognitive intelligence in bridging energy efficiency with cybersecurity.
Relevance of the Paper to the O-CEI Project:
This research paper directly addresses the core objectives of the O-CEI project, specifically contributing to the implementation of Pilot 7: Trustworthy and Secure EV Charging upon Reliable 5G Networks. By proposing a conceptual framework that integrates 5G-enabled IoT microservices with cognitive edge intelligence, this work provides the architectural basis necessary to secure the critical data flows required by Pilot 7. The paper’s focus on trust enhancement and auditable data workflows aligns perfectly with the pilot’s requirement for verifiable and resilient V2G exchanges, ensuring that the expanded attack surface of 5G-connected charging infrastructure is effectively mitigated. Furthermore, the proposed cognitive optimization of power flows supports O-CEI’s broader ambition to demonstrate how the Cloud-Edge-IoT continuum can drive energy efficiency and grid stability in the face of volatile renewable generation.
DriftMoE: A Mixture of Experts Approach to Handle Concept Drifts
Publication: arXiv:2507.18464
Authors: Miguel Aspis,Sebastián A. Cajas,Andrés L. Suárez-Cetrulo,Ricardo Simón Carbajo
Read Abstract
Learning from non-stationary data streams subject to concept drift requires models that can adapt on-the-fly while remaining resource-efficient. Existing adaptive ensemble methods often rely on coarsegrained adaptation mechanisms or simple voting schemes that fail to optimally leverage specialized knowledge. This paper introduces DriftMoE, an online Mixture-of-Experts (MoE) architecture that addresses these limitations through a novel co-training framework. DriftMoE features a compact neural router that is co-trained alongside a pool of incremental Hoeffding tree experts. The key innovation lies in a symbiotic learning loop that enables expert specialization: the router selects the most suitable expert for prediction, the relevant experts update incrementally with the true label, and the router refines its parameters using a multi-hot correctness mask that reinforces every accurate expert. This feedback loop provides the router with a clear training signal while accelerating expert specialization. We evaluate DriftMoE’s performance across nine state-of-the-art data stream learning benchmarks spanning abrupt, gradual, and real-world drifts, testing two distinct configurations: one where experts specialize on data regimes (multi-class variant), and another where they focus on single-class specialization (task-based variant). Our results demonstrate that DriftMoE achieves competitive results with state-of-the-art stream learning adaptive ensembles, offering a principled and efficient approach to concept drift adaptation. All code, data pipelines, and reproducibility scripts are available in our public GitHub repository: https://github.com/miguel-ceadar/drift-moe.
Relevance of the Paper to the O-CEI Project:
The methodology outlined in DriftMoE holds strong potential for adaptation within O-CEI Task 3.5 (Implementation of intra- and cross-domain data management, observability, and AI orchestration mechanisms), where CeADAR plays a leading role. The intent is to use router networks to recommend relevant models from O-CEI’s marketplace. Furthermore, CeADAR is investigating ways to compress a MoE, which would make it even more efficient for recommending resource-intensive models like LLMs, a key part of our future research efforts. The paper acknowledges O-CEI.
Intelligent Edge Computing and Machine Learning: A Survey of Optimization and Applications
Publication: Future Internet, 17(9), 417.
Authors: Sebastián A. Cajas, Jaydeep Samanta, Andrés L. Suárez-Cetrulo, Ricardo Simón Carbajo
Read Abstract
Intelligent edge machine learning has emerged as a paradigm for deploying smart applications across resource-constrained devices in next-generation network infrastructures. This survey addresses the critical challenges of implementing machine learning models on edge devices within distributed network environments, including computational limitations, memory constraints, and energy-efficiency requirements for real-time intelligent inference. We provide a comprehensive analysis of soft computing optimization strategies essential for intelligent edge deployment, systematically examining model compression techniques, including pruning, quantization methods, knowledge distillation, and low-rank decomposition approaches. The survey explores intelligent MLOps frameworks tailored for network edge environments, addressing continuous model adaptation, monitoring under data drift, and federated learning for distributed intelligence while preserving privacy in next-generation networks. Our work covers practical applications across intelligent smart agriculture, energy management, healthcare, and industrial monitoring within network infrastructures, highlighting domain-specific challenges and emerging solutions. We analyse specialized hardware architectures, cloud offloading strategies, and distributed learning approaches that enable intelligent edge computing in heterogeneous network environments. The survey identifies critical research gaps in multimodal model deployment, streaming learning under concept drift, and integration of soft computing techniques with intelligent edge orchestration frameworks for network applications. These gaps directly manifest as open challenges in balancing computational efficiency with model robustness due to limited multimodal optimization techniques, developing sustainable intelligent edge AI systems arising from inadequate streaming learning adaptation, and creating adaptive network applications for dynamic environments resulting from insufficient soft computing integration. This comprehensive roadmap synthesizes current intelligent edge machine learning solutions with emerging soft computing approaches, providing researchers and practitioners with insights for developing next-generation intelligent edge computing systems that leverage machine learning capabilities in distributed network infrastructures.
Relevance of the Paper to the Project:
To achieve the O-CEI project’s goal of accelerating the uptake of innovative Cloud-Edge-IoT solutions and strengthening Europe’s strategic autonomy, a thorough understanding of the current AI landscape is paramount. This review paper provides a necessary up-to-date survey of the state-of-the-art in AI for edge and cloud applications. The knowledge within is foundational for developing some of the project’s core components, including its AIOps solutions and AI-driven marketplace, thereby ensuring that O-CEI is at the forefront of technological advancement.
Offloading Artificial Intelligence Workloads across the Computing Continuum by means of Active Storage Systems
Publication: Future Generation Computer Systems, 2025/12/1, 108271
Authors: Alex Barceló, Sebastián A. Cajas, Jaydeep Samanta, Andrés L. Suárez-Cetrulo, Romila Ghosh, Ricardo Simón Carbajo, Anna Queralt
Read Abstract
The increasing demand for artificial intelligence (AI) workloads across diverse computing environments has driven the need for more efficient data management strategies. Traditional cloud-based architectures struggle to handle the sheer volume and velocity of AI-driven data, leading to inefficiencies in storage, computation, and data movement. This paper explores the integration of active storage systems within the computing continuum to optimize AI workload distribution.
By embedding computation directly into storage architectures, active storage is able to reduce data transfer overhead, enhancing performance and improving resource utilization. Other existing frameworks and architectures offer mechanisms to distribute certain AI processes across distributed environments; however, they lack the flexibility and adaptability that the continuum requires, both regarding the heterogeneity of devices and the rapid-changing algorithms and models being used by domain experts and researchers.
This article proposes a software architecture aimed at seamlessly distributing AI workloads across the computing continuum, and presents its implementation using mainstream Python libraries and dataClay, an active storage platform. The evaluation shows the benefits and trade-offs regarding memory consumption, storage requirements, training times, and execution efficiency across different devices. Experimental results demonstrate that the process of offloading workloads through active storage significantly improves memory efficiency and training speeds while maintaining accuracy. Our findings highlight the potential of active storage to revolutionize AI workload management, making distributed AI deployments more scalable and resource-efficient with a very low entry barrier for domain experts and application developers.
Relevance of the Paper to O-CEI:
This paper performs experiments to offload AI workloads closer to the data sources used to train this model, exploiting near-data locality in the data federation and computing capabilities close to them in the context of MetaOS projects. This is relevant to O-CEI as it uses similar AI tooling, and the experiments occur in the context of the edge-to-cloud continuum, accessing external data sources and computing facilities.
