Reliable Aerial Mobile Communications with RSRP & RSRQ Prediction Models for the Internet of Drones: A Machine...

Area of Science:

• Engineering
• Computer Science
• Telecommunications

Background:

• The unmanned aerial vehicle (UAV) industry is expanding into beyond visual line of sight (BVLOS) operations, requiring robust communication links for applications like environmental monitoring and delivery.
• Terrestrial cellular networks are crucial for enabling BVLOS, but their suitability for aerial use cases needs investigation as they are designed for ground users.
• Ensuring flight safety necessitates a reliable mobile communication link for UAVs operating beyond the operator's sight.

Purpose of the Study:

• To assess the reliability of aerial coverage provided by terrestrial cellular base stations (BSs) for UAV communications.
• To develop and evaluate machine learning models for predicting radio signal strength (RSRP) and quality (RSRQ) in aerial scenarios.
• To understand the impact of distance and elevation angle on cellular signal performance for UAVs.

Main Methods:

• A measurement campaign was conducted using a UAV communicating with a 4G LTE base station in a suburban environment.
• Six machine learning models (multiple linear regression, polynomial, logarithmic) were developed to predict RSRP and RSRQ.
• Models were trained and tested using data as a function of distance and elevation angle between the UAV and BS.

Main Results:

• Terrestrial base stations can provide aerial coverage under specific conditions, influenced by UAV-to-BS distance and flight altitude.
• The proposed machine learning models demonstrated predictive capabilities for RSRP and RSRQ.
• The models achieved root mean square errors (RMSE) of 4.37 dBm for RSRP and 2.71 dB for RSRQ on test data.

Conclusions:

• Terrestrial cellular networks offer potential for supporting BVLOS UAV operations, with performance contingent on geographical and flight parameters.
• Machine learning models provide a viable method for predicting signal reliability in aerial cellular communication scenarios.
• Further research can optimize these models for diverse environments and network technologies to enhance UAV communication.