Work

AIT2-FRL: offline PPO-optimized interval type-2 fuzzy scheduling for fair and sustainable fog–cloud systems

Year: 2026

Type: article

Abstract: Fog–cloud systems support latency-sensitive Internet of Things applications, but heterogeneous and time-varying resources make task scheduling a multi-objective problem. Nodes differ in processing speed, power demand, network delay, failure tendency, and carbon intensity, while tasks vary in urgency and resource requirements. Existing heuristics encode limited trade-offs, population-based optimizers may require repeated search, and direct reinforcement-learning schedulers can obscure deployment-time decisions. This study proposes AIT2–FRL, a two-phase scheduler for independent fog–cloud tasks that combines interval Type-2 fuzzy inference with offline proximal policy optimization. The fuzzy layer represents uncertainty in urgency, congestion, and carbon conditions, establishes a one-time task order, and generates context-dependent weights for makespan, energy, carbon emissions, unreliability, and load imbalance. PPO is used only during offline training to calibrate membership functions and weighting parameters; the validation-selected checkpoint is then frozen. During deployment, feasible nodes are ranked through a deterministic weighted score with linear complexity in the number of candidates. Evaluation covers six workload sizes, 17 matched holdout seeds, four baselines, ablations, sensitivity and convergence analyses, and resource pools up to 256 nodes under common traces and constraints. AIT2–FRL achieved the best mean value for all five primary objectives at every workload, with statistical support at all workloads except unreliability at the smallest size. Secondary response-time and runtime results show that these gains do not extend to every operational measure.

Source: Scientific Reports

DOI: 10.1038/s41598-026-77705-0

Authors: Setareh Moazzami, Abbas Mirzaei, Mehdi Aminian, Ramin Karimi, Nasser Mikaeilvand

Institutions: Islamic Azad University, Tehran, Islamic Azad University North Tehran Branch, Islamic Azad University Central Tehran Branch, Islamic Azad University Ardabil, Islamic Azad University, Shahr-e-Qods Branch

Language: English

Topic: IoT and Edge/Fog Computing

Subfield: Computer Networks and Communications

Field: Computer Science

Domain: Physical Sciences

Open Access status: gold

Estimated APC paid: $2,690

OpenAlex ID: W7224818358