On Analyzing Routing Selection for Aerial Autonomous Vehicles Connected to Mobile Network

Area of Science:

• Computer Science
• Electrical Engineering
• Aerospace Engineering

Background:

• Unmanned Aerial Vehicles (UAVs) require robust communication links for control and data transmission.
• Commercial mobile networks offer a viable, cost-effective solution for UAV connectivity.
• Ensuring reliable data forwarding is critical for UAV mission success, especially in complex scenarios like drone swarms.

Purpose of the Study:

• To propose a novel two-phase algorithm for multi-criteria packet forwarding selection in UAVs.
• To enhance UAV communication reliability by leveraging commercial mobile networks.
• To develop a machine learning-based approach for predicting link behavior during flight.

Main Methods:

• A two-phase algorithm was designed for independent yet constrained data forwarding link selection.
• Machine Learning (ML) models were trained to predict mobile network link performance.
• Implementation involved utilizing three different mobile network operators for redundancy.

Main Results:

• The proposed algorithm effectively manages multi-criteria packet forwarding selection.
• Trained ML models accurately predict link connectivity behavior throughout UAV flight.
• The system demonstrates validated performance in ensuring reliable UAV communication.

Conclusions:

• The developed two-phase algorithm offers an effective solution for UAV packet forwarding.
• Machine learning significantly enhances the prediction of mobile network link stability for UAVs.
• This approach provides a flexible and reliable connectivity framework for various UAV applications, including swarms.