Towards the safe deployment of GNSS-based localization for autonomous trains: A Preliminary Hazard Analysis

Several European projects have investigated Global Navigation Satellite System (GNSS)-based approaches to simplify signalling architectures and reduce the railway infrastructure costs by migrating more functions on board and relying on virtual blocks instead of traditional fixed-block operation. However, introducing GNSS into safety-critical functions creates new failure modes and threat scenarios that challenge existing safety certification practices. This paper presents a first step towards the certification of a GNSSbased localization function through a Preliminary Hazard Analysis (PHA) carried out for autonomous shuttle systems. This PHA identifies GNSS-related events that may lead to hazardous train movements, and provides an initial risk ranking that can guide future design decisions. The analysis combines a functional description of the localization chain and a review of safety-related events from affiliated rail systems (tramways and conventional trains) to compensate the lack of operational data from autonomous shuttles. The PHA highlights a small number of dominant scenarios, including derailment due to erroneous or unavailable GNSS position information. For these scenarios, the paper derives sets of initiating GNSS failures for later quantitative analysis. The scientific contribution of this work is two-fold. First, it proposes a hazard identification approach adapted to the context of autonomous shuttle safety events. Second, it establishes links between GNSS failure modes and systemlevel hazards within an EN 50129-compliant signalling safety case.

Keywords: Satellite positioning system, Railway signalling, Autonomous train, Preliminary Hazard Analysis, Dependability.