Data Center Traffic Prediction Algorithms and Resource Scheduling.

Area of Science:

• Computer Science
• Artificial Intelligence
• Network Engineering

Background:

• Current 5G cloud-network architectures face challenges with resource imbalance and varying demand.
• Accurate network traffic prediction is crucial for efficient resource management and service quality.

Purpose of the Study:

• To develop an intelligent system for predicting network traffic in 5G cloud-network environments.
• To address resource allocation issues and demand differentiation.
• To reduce peak pressure on 5G data center networks.

Main Methods:

• Utilized a time recurrent neural network (a type of Long Short-Term Memory - LSTM model) for traffic prediction.
• Proposed an improved tree species optimization algorithm for initial network data optimization.
• Developed a scheduling algorithm integrating business cooperative caching and load balancing.

Main Results:

• Achieved improved network traffic prediction accuracy using the LSTM model.
• The proposed optimization algorithm enhanced initial data processing.
• The integrated scheduling algorithm effectively reduced peak network load.

Conclusions:

• Intelligent methods, including LSTM and optimized algorithms, significantly improve 5G network traffic prediction.
• The developed scheduling algorithm successfully mitigates peak pressure in 5G data center networks.
• This approach offers a viable solution for managing resources and demand in complex 5G architectures.