Fog Fragment Cooperation on Bandwidth Management Based on Reinforcement Learning

Area of Science:

• Computer Science
• Network Engineering
• Artificial Intelligence

Background:

• The Internet of Things (IoT) generates vast amounts of data, overwhelming network bandwidth.
• Slower bandwidth improvements create bottlenecks, necessitating solutions for efficient data management.
• The proliferation of smart environments and IoT devices increases the demand on network infrastructure.

Purpose of the Study:

• To introduce fog fragment computing as an alternative to conventional fog computing.
• To address bandwidth management challenges in IoT environments, especially for devices with emergencies.
• To develop a cooperative strategy for fog nodes to overcome bandwidth limitations.

Main Methods:

• Formulating the fog node decision-making problem using a reinforcement learning approach.
• Developing a Q-learning algorithm to enable cooperative decision-making among fog nodes.
• Comparing the proposed fog fragment computing method with a single fog node scenario.

Main Results:

• The proposed fog fragment computing method demonstrates superior performance compared to a single fog node approach.
• Cooperative fog nodes effectively manage bandwidth for IoT devices with urgent data needs.
• The Q-learning algorithm facilitates efficient resource allocation and decision-making.

Conclusions:

• Fog fragment computing offers a viable solution for bandwidth constraints in IoT networks.
• Cooperative strategies among fog nodes, guided by reinforcement learning, are crucial for efficient IoT data management.
• This research presents a novel approach to enhance IoT network performance under high data load conditions.