Machine Learning Computing Migration and Management Based on Edge Computing of Multiple Data Sources in the Internet...

Area of Science:

• Computer Science
• Artificial Intelligence
• Network Engineering

Background:

• The proliferation of intelligent terminal devices in smart cities and homes strains traditional computing frameworks.
• Increasing data volumes and computational demands necessitate advanced solutions beyond conventional architectures.

Purpose of the Study:

• To research the migration and management of deep reinforcement learning (DRL) computing within Internet of Things (IoT) edge computing environments.
• To integrate DRL for optimizing edge computing migration schemes and resource allocation management.

Main Methods:

• Integration of deep reinforcement learning algorithms into IoT edge computing frameworks.
• Development of optimized schemes for computing task migration and resource allocation.
• Comparative analysis against traditional algorithms and minimum migration strategies.

Main Results:

• DRL effectively controls computing migration costs and ensures efficient task completion with stable operation.
• The proposed management model demonstrates reduced energy consumption compared to traditional methods.
• Shorter average computing waiting times were achieved with the DRL-based approach.

Conclusions:

• DRL-based edge computing offers a viable solution for the growing computational demands of intelligent devices.
• Optimized migration and resource management through DRL enhance the efficiency and reduce the energy footprint of IoT systems.
• The findings support the adoption of DRL for improving edge computing performance in smart environments.