Video coding deep learning-based modeling for long life video streaming over next network generation

Area of Science:

• Computer Science
• Electrical Engineering

Background:

• Smart network availability is crucial, especially during high video streaming traffic.
• Video streaming systems require efficient resource management for optimal performance.

Purpose of the Study:

• To propose a deep learning methodology for predicting video streaming quality, power consumption, and bandwidth.
• To enhance the availability of video streaming systems in smart networks.

Main Methods:

• Developed a deep learning prediction model for video streaming metrics.
• Utilized H.264/AVC codec parameters (resolution, quantization) as input.
• Trained, validated, and tested models using extensive experimental data.

Main Results:

• Accurate prediction models were built for video streaming quality, power consumption, and bandwidth.
• The models demonstrated high accuracy in predicting key video streaming parameters.
• The methodology proved effective in enhancing network availability.

Conclusions:

• Deep learning models can accurately predict video streaming quality, power, and bandwidth.
• These predictions are valuable for improving network availability in cooperative environments.
• The proposed methodology offers a robust solution for managing video streaming resources.