Intelligent Traffic Adaptive Resource Allocation for Edge Computing-based 5G Networks

Area of Science:

• Computer Science
• Telecommunications Engineering
• Artificial Intelligence

Background:

• Increasing mobile traffic strains fifth-generation (5G) networks, particularly for ultra-high reliability and ultra-low latency (uRLLC) services.
• Existing 5G, edge computing, and IoT-Cloud integrations face challenges in meeting stringent uRLLC requirements.
• Data-driven methods are crucial for managing mobile traffic but require advanced AI solutions.

Purpose of the Study:

• To develop and evaluate an AI-driven approach for controlling mobile traffic flow in 5G networks.
• To improve the reliability and reduce latency for uRLLC communication scenarios.
• To present an intelligent architecture for dynamic resource dispatching in multi-site environments.

Main Methods:

• Developed a traffic-flow prediction algorithm using Long Short-Term Memory (LSTM) with an attention mechanism for single-site traffic data.
• Proposed an intelligent IoT-based mobile traffic prediction-and-control architecture for multi-site resource management.
• Employed experimental evaluations to demonstrate the effectiveness of the proposed AI scheme.

Main Results:

• The LSTM-based algorithm accurately predicts peak mobile traffic flow values.
• The proposed architecture effectively reduces communication latency.
• The AI scheme significantly lowers the packet-loss ratio, enhancing communication reliability.

Conclusions:

• Artificial intelligence, specifically LSTM with attention, is effective for mobile traffic flow prediction and control in 5G networks.
• The intelligent IoT-based architecture enables dynamic resource dispatching, crucial for uRLLC.
• The study demonstrates a viable AI solution for optimizing 5G network performance and meeting demanding communication requirements.