Trajectory Prediction Model for Accurate Navigation and Surveillance Using kNN Regressor

Accurate trajectory prediction is critical for airspace surveillance and navigation especially in GPS-denied environments. This paper explores the use of a non-parametric Machine Learning (ML) model to estimate aviation traffic trajectories. Existing approaches lack accuracy and do not consider the effect of Automatic Dependent Surveillance-Broadcast (ADS-B)/Global Positioning System (GPS) dropouts and achieve locational accuracies of up to 154 meters in 2D (latitude and longitude) estimations. Additionally, multi-path fading, and ADS-B/GPS interferences may further impede trajectory prediction accuracies. In this work, the k-nearest Neighbors (kNN) Regressor is used to improve trajectory estimations under a GPS interference event. Our results show improvement, achieving an accuracy of 94 m for 2D latitude and longitude estimations. Such improvements in accuracy are critical for optimal trajectory estimations in GPS-denied environments.