Federated learning enables intelligent reflecting surface in fog-cloud enabled cellular network.

Area of Science:

• Wireless Communication Systems
• Artificial Intelligence
• Signal Processing

Background:

• Intelligent Reflecting Surface (IRS) technology reconfigures wireless environments using numerous small reflecting units.
• Existing IRS mechanisms centralize data learning and decision-making, overlooking energy-efficient and delay-aware learning.
• Decentralized learning approaches for IRS-assisted communication remain largely unexplored.

Purpose of the Study:

• To propose a novel algorithm for energy-efficient and delay-aware IRS-assisted communication.
• To enhance high-speed data transmission through optimized offloading and scheduling in IRS systems.
• To address the limitations of centralized learning in IRS by introducing a federated learning approach.

Main Methods:

• Development of the Federated Learning aware Intelligent Reconfigurable Surface Task Scheduling (FL-IRSTS) algorithm.
• Distributed model training across multiple nodes within a local healthcare fog-cloud network.
• Generation of a global model by aggregating locally trained models for optimized IRS configuration.

Main Results:

• The FL-IRSTS algorithm achieves high-speed communication with improved energy and delay efficiency.
• Local processing of healthcare data during federated learning ensures privacy and reduces latency.
• Simulation results demonstrate that the proposed algorithm approaches the performance of centralized machine learning (ML).

Conclusions:

• The FL-IRSTS algorithm offers an effective solution for energy and delay-efficient IRS-assisted wireless communication.
• Federated learning enables decentralized decision-making and model training in IRS systems.
• The proposed approach balances communication rate goals with energy and delay constraints.