Edge-to-Cloud Computations-as-a-Service in Software-Defined Energy Networks for Smart Grids

TL;DR

This work introduces Edge-to-Cloud Computations-as-a-Service within Software-Defined Energy Networks (SDEN) to address real-time analytics needs in smart grids by co-optimizing computation placement and networking across edge, fog, and cloud layers under URLLC constraints. It formalizes a joint energy–latency offloading framework, provides an analytical solution under simplifying assumptions and a scalable greedy heuristic, and develops a tiered AI pipeline with federated Graph Neural Networks for privacy-preserving fault detection and microgrid coordination. The results demonstrate substantial gains in energy efficiency, reduced bandwidth, lower latency and jitter, near-absolute availability, and accurate distributed fault detection, validating the viability of a programmable, resilient CaaS substrate for grid-scale operations. The findings suggest significant practical impact for grid resilience, DER/DR coordination, and privacy-aware management, with future work including hardware-in-the-loop validation and interoperability with existing grid management systems.

Abstract

Modern power grids face an acute mismatch between where data is generated and where it can be processed: protection relays, EV (Electric Vehicle) charging, and distributed renewables demand millisecond analytics at the edge, while energy-hungry workloads often sit in distant clouds leading to missed real-time deadlines and wasted power. We address this by proposing, to our knowledge, the first-ever SDEN (Software Defined Energy Network) for CaaS (Computations-as-a-Service) that unifies edge, fog, and cloud compute with 5G URLLC (Ultra-Reliable Low-Latency Communications), SDN (Software Defined Networking), and NFV (Network Functions Virtualization) to co-optimize energy, latency, and reliability end-to-end. Our contributions are threefold: (i) a joint task offloading formulation that couples computation placement with network capacity under explicit URLLC constraints; (ii) a feasibility preserving, lightweight greedy heuristic that scales while closely tracking optimal energy and latency trade-offs; and (iii) a tiered AI (Artificial Intelligence) pipeline-reactive at the edge, predictive in the fog, strategic in the cloud-featuring privacy-preserving, federated GNNs (Graph Neural Networks) for fault detection and microgrid coordination. Unlike prior edge-only or cloud-only schemes, SDEN turns fragmented grid compute into a single, programmable substrate that delivers dependable, energy-aware, real time analytics establishing a first-ever, software defined path to practical, grid-scale CaaS.

Figures (11)

• Figure 1: A Conceptual Representation of SDEN for CaaS
• Figure 2: Task assignment ratios and node utilisation levels across edge, fog, and cloud tiers under greedy heuristic vs. round-robin allocation for $N=50$ tasks. Simulation setup: $\beta_i \in [0.67,2]$ tasks/s, $\mu_{\text{edge}}=5$, $\mu_{\text{fog}}=15$, $\mu_{\text{cloud}}=60$ tasks/s, $\eta_{\text{proc}}=\{0.2,0.5,2.0\}$ J/bit, $\eta_{\text{comm}}=\{0.1,0.3,1.2\}$ J/bit, $\omega_1=0.6$, $\omega_2=0.4$.
• Figure 3: Bandwidth utilisation (Mbps) across cloud-only, edge-first, and greedy heuristic allocation models for $N=50$ tasks. Each task requires 1 MB/s bandwidth; preprocessing reduces transmission by $90\%$ at edge and $50\%$ at fog. Simulation setup: $\beta_i \in [0.67,2]$ tasks/s, $\mu_{\text{edge}}=5$, $\mu_{\text{fog}}=15$, $\mu_{\text{cloud}}=60$.
• Figure 4: Comparison of total energy consumption (kWh) between cloud-only and hybrid edge–fog–cloud models under greedy heuristic allocation. Results averaged over 1000 trials with random seeds; maximum saving $69.65\%$, median saving $30.2\%$. Simulation setup: $N=50$ tasks, $\beta_i \in [0.67,2]$ tasks/s, $\eta_{\text{proc}}=\{0.2,0.5,2.0\}$ J/bit, $\eta_{\text{comm}}=\{0.1,0.3,1.2\}$ J/bit, $\omega_1=0.6$, $\omega_2=0.4$.
• Figure 5: End-to-end latency distribution under greedy heuristic allocation with URLLC enabled. Transmission delay reduced by $15\%$ and jitter by $25\%$ relative to baseline. Simulation setup: varying workloads up to $N=100$ tasks, $\beta_i \in [0.67,2]$ tasks/s, $\mu_{\text{edge}}=5$, $\mu_{\text{fog}}=15$, $\mu_{\text{cloud}}=60$, $\omega_1=0.6$, $\omega_2=0.4$.