همایش بین المللی هوش مصنوعی

صفحه اصلی / دومین همایش بین المللی هوش مصنوعی

Hybrid Fuzzy‑Q Learning Approach for Load Balancing in IoT Networks

نویسندگان :

Fereshteh Taghizadeh¹ Mohsen Raji² Morteza Keshtkaran³

1- دانشگاه شیراز 2- دانشگاه شیراز 3- دانشگاه شیراز

کلمات کلیدی :

RPL،Load Balancing،Fuzzy Logic،Q-learning،IoT Networks،Parent Selection،Reinforcement Learning

چکیده :

Artificial Intelligence is redefining the optimization of complex IoT routing systems, enabling adaptive, data‑driven decision‑making in scenarios where traditional methods fall short. In traditional IoT networks, maintaining reliable and energy‑efficient routing under dynamic traffic patterns and topology changes remains a critical challenge. To address this, we propose Fuzzy‑QRPL, a hybrid load‑balancing framework that integrates fuzzy logic with Q‑learning to enhance routing efficiency in RPL-based IoT networks. Two key metrics—Residual Energy Potential (EP) and Congestion Index (CI)—are computed from queue occupancy, end‑to‑end delay, number of child nodes, and residual energy. Nine fuzzy rules are defined, each initialized with a Q‑value and updated according to a reward based on the packet delivery ratio (PDR). This AI‑enhanced mechanism enables dynamic and intelligent parent selection under varying network conditions. Cooja simulation results across diverse network sizes demonstrate that Fuzzy‑QRPL significantly improves PDR, reduces delay, and increases the number of successfully delivered packets compared to conventional RPL objective functions.