Target Classification Based On Kinematic Data From AIS/ADS-B, Using Statistical Features Extraction and Boosting
Raphael Ginoulhac, Frédéric Barbaresco, Jean-Yves Schneider, Jean-Marie Pannier, Sebastien Savary · 2019
We propose a simple yet efficient method to classify multivariate time series with an arbitrary number of timesteps, and we apply it to the classification of targets (either aircrafts or vessels) using kinematic data only. We use data obtained from the Automatic Identification System (AIS) and the Automatic Dependent Surveillance-Broadcast (ADS-B) to get labelled trajectories for supervised learning, as a proof of concept for later use on radar tracks. The method consists in extracting statistical features from each temporal variable (speed, acceleration, etc.), and then feeding them to a Gradient Boosting classifier. We show that the performance of this method is on par with the state of the art, as the classification accuracy is close to 86AIS data, and thus that it could be used in radars for the classification of targets based on their trajectory.