Multi-Criteria Device Scheduling in Blockchain-Assisted Federated Learning Using Pythagorean Fuzzy Sets and Ordered Weighted Averaging
No Thumbnail Available
Date
2026
Journal Title
Journal ISSN
Volume Title
Publisher
Library Information Services, COMSATS University Islamabad, Lahore Campus
Abstract
Federated Learning (FL) in Internet-of-Things (IoT) environments faces challenges such as
device heterogeneity, energy limitations, and unfair client participation. Existing classical
decision-based client selection methodologies like TOPSIS and VIKOR are based on crisp
dominance relations and, as such, would repeatedly give preference to high-end devices,
resulting in accelerated battery depletion, reward imbalance, and reduced participation fair-
ness. To address these limitations, this thesis proposes a workload-aware fuzzy scheduling
framework based on Pythagorean Fuzzy Sets (PFS) and the Ordered Weighted Averag-
ing (OWA) operator within a blockchain-assisted FL environment. The proposed model
incorporates multi-dimensional criteria to include energy cost, dataset contribution, CPU
capability, deadline feasibility, reputation, and income deficit to enable adaptive decision-
making under uncertainty under different workload intensities.
A discrete-event simulation of 60 heterogeneous edge devices over 150 global rounds
demonstrates that the proposed PFS with OWA scheduler achieves an improved trade-off
between accuracy, energy usage, and fairness compared with a deterministic TOPSIS–
VIKOR. The Lorenz-curve analysis indicates the superior equity of incentive distribution
with PFS with OWA.
These results demonstrate that accuracy-only scheduling is insufficient for realistic
large-scale FL systems, where fairness, energy preservation, and device longevity are equally
critical. Therefore, the proposed PFS with OWA framework provides an adaptive, inter-
pretable, and resource-aware scheduling solution for next-generation blockchain-integrated
federated learning systems.
Description
Keywords
Department of Mathematics, Mathematics, FA24, Multi-Criteria Device Scheduling, Blockchain-Assisted Federated Learning, Pythagorean Fuzzy Sets, Dr. Sarfraz Ahmad
