Filtering Techniques for ADS-B Trajectory Preprocessing

Abstract

This paper addresses the issue of noisy, uncertain and quantized data in crowdsourced ADS-B and Mode S data and explores propositions of implementations of preprocessing techniques to address them. After a description of ADS-B data focused on sources of noise and uncertainty, we present in detail a selection of filters that have been implemented in the traffic library, and widely used in the constitution of open datasets used in further research. We also illustrate the results of the filtering with trajectory data collected by The OpenSky Network and by inexpensive RTL-SDR receivers.