Predictive Modeling of Multilink Delay in Avionic Networks: a Machine Learning Approach for Enhanced Communication Reliability

The rapid growth of global air traffic necessitates efficient communication systems to ensure the safety and reliability of modern Air Traffic Management. Avionic networks, equipped with multiple data links, play a pivotal role in enabling robust communication between aircraft and ground systems. However, ensuring seamless transitions across heterogeneous networks while minimizing delays and maintaining Quality of Service remains a significant challenge. This paper presents a machine learning-based predictive modeling approach to address these challenges by forecasting end-to-end communication delays in avionic networks. Synthetic data, generated using a system-level simulator based on the OMNeT++ framework, captures key aspects of avionic communication, including aircraft mobility, satellite and terrestrial data links, communication delays, and signal quality. Among the four machine learning models evaluated, the Random Forest model emerged as the most accurate, achieving the lowest Normalized Root Mean Square Error. The proposed methodology successfully predicted overload scenarios, enabling proactive switching to satellite backup links and significantly improving overall network performance. By reducing delays and ensuring consistent Quality of Service throughout flight operations, this approach enhances the reliability and efficiency of communication systems. These findings highlight the transformative potential of machine learning-based predictive modeling in advancing the efficiency, scalability, and decision-making capabilities of avionic communication systems, paving the way for safer and more efficient air traffic management.