Handover for V2V communication in 5G using convolutional neural networks

Area of Science:

• Telecommunications Engineering
• Artificial Intelligence
• Computer Vision

Background:

• Vehicle communication is crucial for traffic flow and safety.
• 5G technology offers high data rates and quality of service for advanced transportation systems.
• Deep learning (DL) excels at processing large datasets for feature identification.

Purpose of the Study:

• To detect vehicles and identify obstacles using DL in 5G environments.
• To develop a novel horizontal handover prediction method for heterogeneous networks.
• To improve vehicle communication efficiency and safety.

Main Methods:

• Utilized the VGG19 deep learning model via transfer learning for vehicle and obstacle detection.
• Proposed a new horizontal handover prediction algorithm based on channel characteristics.
• Implemented and evaluated algorithms within simulated 5G network environments.

Main Results:

• Achieved a 97% success rate in identifying vehicles.
• Successfully predicted the next station for horizontal handover.
• Demonstrated the effectiveness of DL algorithms in 5G vehicle communication.

Conclusions:

• The proposed DL-based vehicle detection and handover prediction methods are effective in 5G environments.
• The VGG19 model and novel handover algorithm show high accuracy and reliability.
• This research contributes to safer and more efficient intelligent transportation systems.