Framework to Classify Railway Track Areas According to Local GNSS Threats

Abstract

In this paper we present a modular framework to classify railway track areas regarding the expected presence of local GNSS threats. This information might be critical for a safe signalling operation, for example to determine where virtual balises could be placed safely. We show first how different GNSS threats can be detected using dedicated detection algorithms and how these individual detection results can be then transformed from time to the track domain. An overall decision logic is subsequently used to identify an area as suitable or unsuitable for GNSS usage by combining all available GNSS data collected over the same track area. Finally, the framework implementation is evaluated with railway data obtained during a measurement campaign in Sardinia, Italy in 2019. Even though developed in the railway context, the presented framework architecture and methodology may be also considered to perform similar classification tasks for other means of transport.