Video Quality Control Method Based on Mobile Network Quality Measurements and Predictions for Reliable Remote Monitoring in Autonomous Driving

Autonomous driving on public roads requires remote monitoring systems for safety. To realize these systems, researchers in academia and industry have focused on controlling network quality metrics, such as bandwidth or latency, and application quality metrics, such as bitrate and framerate. Though conventional application control techniques are based on measured network quality metrics, this causes application deterioration due to late control when the network quality metrics change rapidly. Therefore, recent research has shown the effectiveness of the application control methods based on predicted network quality metrics. However, these methods have limited prediction accuracy, and the predictions might lead to wrong values and critical system problems such as video stalls. To address this issue, we propose a bitrate control method combining both measured and predicted network quality metrics simultaneously to use them complementarily. Finally, in a field experiment, we show our method achieves 28% fewer video stalls and 10% shorter total video stall time than prior methods. This method should contribute to providing reliable remote monitoring services for mobile devices such as autonomous vehicles, machines, and drones.