Work

A Cyber-Resilient Edge–Fog–Cloud Framework for Forecast-Driven Energy Management in Networked Microgrids

Year: 2026
Type: article
Abstract: The increasing penetration of renewable energy resources and the interconnection of microgrids necessitate an energy management framework that provides robust optimization, scalability, and resilience against cyber threats. Cyber threats, such as False Data Injection Attacks (FDIAs), can compromise the integrity of sensors, leading to suboptimal decision-making. Therefore, this paper proposes a cyber-resilient assisted Edge–Fog–Cloud framework for optimal energy management in Networked Microgrids(NMGs). The proposed architecture utilizes a cloud layer for model training, while real-time forecasting, FDIAs detection and mitigation, and local microgrid operation optimization take place at the edge layer. Additionally, coordinated inter-microgrid energy management uses a fog layer. At the edge layer, a Gated Recurrent Unit (GRU) network, identified as the best-performing option, is employed for photovoltaic (PV) and load forecasting to support proactive scheduling. The edge and fog layer also employ a bi-level Mixed-Integer Nonlinear Programming (MINLP) formulation that governs both local microgrid operation and network-level energy exchange using forecasted data, respectively. Additionally, the FDIAs detection and mitigation module is also present at the edge for real-time validation of PV and load values. To simulate cyberattacks in NMGs, a Fault Injection and Detection Analysis (FIDA) framework is employed to synthesize representative FDIAs, including stuck-at, zero-out, and noise-based stealth attacks. Simulation results using real-world data demonstrate that the proposed framework preserves forecasting accuracy within an acceptable uncertainty interval, maintains the benefits of bi-level optimization, and ensures secure coordinated operation under cyberattacks, achieving an average mitigation error of 1.06 kW, while keeping load and PV measurements within their admissible uncertainty bounds of 3.39 kW and 0.59 kW, respectively, in the worst case.
Language: English
Open Access status: hybrid
Estimated APC paid: $2,645
OpenAlex ID: W7214939619
FWCI:
Cites:
Cited by:
Related to: 10