Sustainable Resource Allocation and Reduce Latency Based on Federated-Learning-Enabled Digital Twin in IoT Devices

Area of Science:

• Computer Science
• Network Engineering
• Artificial Intelligence

Background:

• Edge computing and IoT generate vast data, requiring efficient processing.
• Current systems face challenges in data privacy, security, and scalability.
• Digital Twins (DT) offer a virtual representation for system monitoring and control.

Purpose of the Study:

• To develop a secure and scalable edge network solution using Digital Twins (DT) and blockchain.
• To propose a Federated Learning (FL) framework integrated with DT for enhanced data privacy and system reliability.
• To optimize resource allocation (RA) and minimize energy consumption (EC) for real-time IoT data processing.

Main Methods:

• Utilized Digital Twins (DT) integrated with edge networks and blockchain technology.
• Developed a Federated Learning (FL) framework running on a blockchain, powered by the DT edge network.
• Implemented a DT-empowered Deep Reinforcement Learning (Deep-RL) agent for sustainable Resource Allocation (RA).

Main Results:

• The proposed DT-empowered Deep-RL agent effectively balances system latency and Energy Consumption (EC).
• The framework demonstrated enhanced data privacy, system security, and reliability.
• Simulation results showed the DT-based approach could perform 47.5% of computing activities locally with 1 MHz bandwidth, minimizing transmission costs.

Conclusions:

• The integrated DT, blockchain, and FL framework provides a robust solution for edge computing challenges.
• The Deep-RL agent optimizes resource distribution, leading to improved performance and efficiency in IoT environments.
• This approach enhances the security, privacy, and real-time processing capabilities of edge networks.