Cyber-Resilience in Future Buildings: AI-Driven Cloud Security for Sustainable Energy Management Systems
Journal of Future Building · 17.06.2026 · 3 (2) · 1-11
DOI: 10.66320/1garrr09 ↗Kurz gefasst
Fachartikel in der Zeitschrift Journal of Future Building, online erschienen am 17.06.2026. Verfasst von Adeel Ali. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Net-zero building strategies increasingly depend on cloud‑connected Internet of Things (IoT) sensors and Energy Management Systems (EMS) to minimize carbon footprints, yet this deep integration exposes building physics to cyber‑physical attacks that can invert sustainability gains. Compromised HVAC and smart grid interfaces can trigger massive energy waste instead of conservation, yet existing network intrusion detection systems lack the physical context to distinguish malicious control commands from legitimate energy‑ optimization fluctuations. This study proposes and empirically validates a Cyber‑Resilient AI Architecture (CRA‑EMS) that couples unsupervised deep anomaly detection with automated system isolation to guarantee energy sustainability during active cyber intrusions. The framework deploys a Temporal‑Graph Variational Autoencoder (TG‑VAE) and contrastive self‑supervised learning at the cloud edge gateway, learning normative telemetry patterns from a digital twin of a 10,000 m² medium office building. Anomaly thresholds are established via Peaks‑Over‑Threshold (POT) extreme value theory to maintain a stringent 0.05 false‑positive rate per day; upon detection, the gateway isolates the compromised HVAC subnet and activates fail‑safe schedules. Co‑simulation with EnergyPlus 23.2 and ns‑3.38 quantifies energy deviation across fifteen realistic attack scenarios. The TG‑VAE achieves an Area Under the ROC Curve (AUC) of 0.984 and a recall of 97.1 % at the target false‑positive rate, significantly outperforming baseline LSTM‑VAE and Isolation Forest models. Energy waste during a two‑hour chiller setpoint attack is reduced by 98.0 % (from 174.2 kWh to 3.4 kWh), and attack dwell time is compressed from hours to under 30 seconds. These findings establish cyber‑resilience as an inseparable component of sustainable building certification, offering civil engineers and cloud architects a quantifiable, mathematically rigorous blueprint for secure, net‑zero infrastructure.
Zitieren
APA
Ali, A. (2026). Cyber-Resilience in Future Buildings: AI-Driven Cloud Security for Sustainable Energy Management Systems. Journal of Future Building, 3(2), 1-11. https://doi.org/10.66320/1garrr09
Vancouver
Ali A. Cyber-Resilience in Future Buildings: AI-Driven Cloud Security for Sustainable Energy Management Systems. Journal of Future Building. 2026;3(2):1-11. doi:10.66320/1garrr09
BibTeX
@article{Ali2026,
author = {Ali, Adeel},
title = {Cyber-Resilience in Future Buildings: AI-Driven Cloud Security for Sustainable Energy Management Systems},
journal = {Journal of Future Building},
year = {2026},
volume = {3},
number = {2},
pages = {1--11},
doi = {10.66320/1garrr09}
}
EUVIMED