Satellite positioning set to enhance train localisation

As Europe’s railways expand, their safety and efficiency depend increasingly on knowing every train’s position. Traditional train detection systems rely on devices mounted along the rails, such as track circuits and axle counters. Trains also communicate their track positions and distance travelled, thanks to Eurobalises and odometers.

However, the focus now is on achieving even greater accuracy and real-time localisation of trains by improving onboard positioning technology. This will also allow the removal of trackside detection elements, which is expected to positively impact the overall Life Cycle Cost (LCC) and in particular to reduce the need for maintenance. One solution is to look up to the skies and harness Global Navigation Satellite Systems (GNSS), e.g. the EU’s Galileo, as part of the further evolution of the European Train Control System (ETCS) specifications.

Under EU-Rail, building on the work of its predecessor programme Shift2Rail, researchers explored ways in which satellite technologies can be used to pinpoint the position of trains with high accuracy.

Six technical demonstrators (TDs) were completed on the proposed concept of Fail-Safe Train Positioning (FSTP), looking at the application of GNSS and different augmentation technologies, as well as other onboard existing sensors. The lab and on-site demos highlighted how Satellite-Based FSTP has great potential to improve the localisation of trains.

Roadmap and Migration Strategy

The first goal was to define an interoperable solution for Satellite-Based FSTP for railways, calling on two approaches: a Virtual Balise (VB) concept and a Stand-Alone FSTP system.

Research partners produced a Roadmap and Migration Strategy. This identified the need to find possible commonalities and synergies between the two above-mentioned approaches – how to fully include fail-safe and interoperable train positioning solutions in future Technical Specifications for Interoperability (TSIs), and a Cost Benefits and Impact Analysis for the analysed solutions. The strategy also noted that the concept of Satellite-Based FSTP systems should be integrated in current and future Control-Command and Signalling (CCS) systems.

Virtual Balise-based Train Positioning

In the VB concept, positioning information is sent to a train through GNSS signals, supported by an ‘augmentation network’ (extra sensors), instead of relying only on physical balises along the track. Here, the train’s onboard equipment includes a new module – the Virtual Balise Reader, which must be integrated into the existing European Train Control System or ETCS onboard architecture. This reader processes the GNSS signals and compares the GNSS coordinates with the list of coordinates onboard. Next, the reader reports the corresponding virtual balise to the Eurocab (a train-based computer), when the coordinates stored for it are reached.

Three demos were completed with the aim of proving the feasibility of the VB detection, by mixing the latest satellite positioning technologies, an augmentation network and kinematic sensor technologies:

Demo: Simplified test bench to support development of a Virtual Balise demonstrator

• Solution: VB technology through GNSS to enhance positioning information for a train.
• TRL: 5 (technology validated in a relevant environment).
• Who benefits: Infrastructure managers, Railway operators.

Discussion

Field tests were completed in Czechia on a basic maintenance track vehicle (MUV 69). As this train did not allow installation of all the initially planned sensors, it was simplified by replacing the speed sensors (wheel encoders, Doppler radar) with a GNSS receiver that came with its own interface. Because the laboratory testbed used real data from a measurement campaign in a railway environment, there was no need to simulate any environmental impacts (e.g. hills, tunnels, etc.) on GNSS signals, only those already recorded in the input dataset.

Key findings

In the demo of VBs in Czechia, the position errors were mostly around a few metres, and in only one case did the error exceed 10 metres. So incorporating track data into the onboard algorithm can be a crucial factor in enhancing safety and accuracy.

The lab tests and results provided valuable feedback for the demo developers, by verifying that the equipment operated correctly and by adjusting the algorithm parameters. Researchers now have a better idea of what can be further developed for any solution that will use GNSS for estimating a train’s safe position. The key outcome was an evaluation of the performance characteristics for each of the individual internal functions. KPIs (Key Performance Indicators) were set for different aspects, including for position accuracy, speed accuracy, and protection level.

Demo: FSTP solution applying the Virtual Balise concept in a real ERTMS/ETCS environment for regional low-traffic lines

• Solution: VB can be effectively used in FSTP, within an ERTMS/ETCS environment, on non-urban rail lines.
• TRL: 5 (technology validated in a relevant environment).
• Who benefits: Infrastructure managers, Railway operators.

Discussion

If secondary rail lines are to survive and thrive in future, the regional low-traffic ERTMS/ETCS system must evolve significantly. This system must meet key requirements such as safety, reliability, availability, health, environmental protection, technical compatibility and reduced costs. So this VB demo set out to prove the effectiveness of an FSTP solution that uses the VB concept in a real ERTMS/ETCS environment, and in this case on regional low-traffic rail lines.

A key feature of the demo was fully integrating the VB concept with ERTMS/ETCS, to highlight the feasibility of a new technology (satellite-based positioning) in a well-established operational context.

To enhance the satellite positioning of the VBs, the demo used the current version of the European Geostationary Navigation Overlay Service (EGNOS), Europe’s regional Satellite-based Augmentation System. This is because a fully suitable EGNOS version for railway applications (capable of handling railway safety certification processes as well as supporting multi-frequency and multi-constellation use) is not yet available. Initial results will be made in available by 2027, once the first phase of demonstrations in Europe’s Rail has been completed, using an already available version initially developed for the aviation sector.

To enable comparisons between positioning applications using EGNOS and others that implement a dedicated augmentation network, an ‘alternative’ Local Area Augmentation (LCA) network architecture was developed. So the demo consisted of two separate hardware products: one using current EGNOS and the other employing the LCA network.

The demo’s main goal was the full integration of the VB concept into the ERTMS/ETCS Level 2 Architecture, in order to:

• Assemble and install the integrated system, thus the total combination of subsystems and components required to form the complete system.
• Demonstrate that the integrated system, subsystems and components work together as defined by the interfaces.
• Demonstrate that the integrated system, subsystems and components meet RAMS (Reliability, Availability, Maintainability & Safety) requirements for interlocking systems.

The pilot line was an ERTMS/ETCS-equipped commercial line (Novara-Rho, a line operated by RFI) in Italy, using a diesel locomotive already equipped with the latest version of the European Vital Computer (EVC).

A lab was set up with full hardware and software instruments. This testbed aimed to enable testing of equipment under both nominal and extreme conditions, including support for fault-injection testing. The testbed complemented field testing.

Key findings

This demo in Italy successfully tested a local augmentation network and EGNOS, in a fully integrated ETCS (European Train Controlling System), both onboard and trackside operating on a real line. Position error with local augmentation was below five metres; with EGNOS the error is usually around five metres, but with a higher variance.

The demo tests successfully demonstrated the feasibility of integrating satellite technology with standard systems for positioning, without compromising interoperability – this is essential for full integration into the ERTMS/ETCS Level 2 Architecture. Integration tests also confirmed that the system, when properly assembled and installed with its subsystems and components, meets the RAMS (Reliability, Availability, Maintainability and Safety) requirements.

The Novara-Rho line was an essential testing ground, featuring diverse environments: rural, urban, and tree canyons, including bridges, overpasses, and tunnels. This diversity enabled rigorous testing of the positioning algorithm and led to excellent results. Testing of the current version of EGNOS highlighted the environmental challenges for train positioning, suggesting that enhancements to come in EGNOS v3 will be beneficial.

The system was tested and analysed according to a System Integration Planning. This showed that all the system’s subsystems and components interact correctly and perform the functions identified in the Test Specification.

Demo: FSTP solution applying the Virtual Balise concept in a real ERTMS/ETCS environment: testing interoperability between regional and main railway lines

• Solution: Ensuring interoperability between regional and main railway lines, for an FSTP solution plus VBs in a real ERTMS/ETCS environment.
• TRL: 4 (technology validated in a lab).
• Who benefits: Infrastructure managers, Railway operators.

 Discussion

Given the necessary technological evolution of Regional Low-Traffic lines and the strong requirement for ERTMS-based interoperability between regional and main lines, this demo also aimed to test FSTP implementing the concept of VB. The goal was to compare and analyse train positions calculated through a VB device with those positions calculated by an independent physical balise set.

The demo included laboratory tests using real data acquired in the field, carried out on the RFI Bari-Brindisi line in Italy. This line is equipped with Italy’s own SCMT signalling system (similar to ETCS level 1, with Eurobalises).

To ensure a physically verifiable position along the line, a group of VBs were linked to a group of real balises. This allowed comparison of the difference (timing errors) between the positions of the physical balises with those of the virtual ones.

A simple test vehicle was fitted with the GNSS receiver, 3D Inertial Measurement Unit (including accelerometers and gyroscopes) and the odometer (wheel encoder). The data recorded onboard the train were used in the lab to evaluate the deviation of the VBs, compared with the real balises.

Key findings

The lab tests, which used data recorded from several field trips, showed that the implemented algorithm is robust enough to cope with environmental disturbances to train positioning. However, for a full validation of the proposed solution, further runs on different lines will be necessary to verify different environments and to use different sets of visible satellites.

Stand-Alone Fail-Safe Train Positioning

Earlier studies, focusing just on GNSS, showed that this satellite technology alone cannot guarantee Stand-Alone Fail-Safe Train Positioning (estimation of a train’s position and speed), either for performance or for safety-critical applications. So researchers envisage adding a Safe Fusion Algorithm (SFA) functional block. This module can combine GNSS signals with data from other sensors, which could result in a more accurate and safe odometry (distance calculation subsystem).

In this second approach (Stream 2), the goal was to finalise the activities to support the development of Stand-Alone FSTP prototypes. These were tested on three demos, covering multiple track lines and environmental scenarios, with trains running in France, Germany and Spain.

All three prototypes were based on a module known as the ‘macro-block’ (E_ODO-OB) to estimate the position and speed, and to report balise telegrams (line data messages) to be delivered to ETCS onboard the train (ETCS-OB). The block receives Enhanced Odometry from Track Side (E_ODO-TS) and includes four functional blocks: Safe Fusion Algorithm (SFA); Balise Telegram Reporter; Localisation Sensors which can collect information from GNSS (or a balise reader), Accelerometer, Gyroscope and Wheel Angular Speed Sensor; and Data Client Manager.

After feedback from the demos, changes were made to the original System Architecture Specification for Stand-Alone FSTP. Only the interfaces between the E_ODO-OB subsystem, and both the ETCS-OB and E_ODO-TS, were standardised.

The algorithms proposed in the three prototypes are different, but they all comply with the architecture inputs previously defined. Each demo also had a different focus within the FSTP algorithm, in order to demonstrate different parts of the whole system.

Key findings

The three different demos of the Stand-Alone FSTP system highlighted the feasibility of GNSS-based algorithms for safe positioning:

• Solution 1 tested in Spain: this real-time demo (integrated GNSS, IMU, tachometer and a Digital Map) with an algorithm showed it is possible to have a positioning system of around 20 metres fixed error.
• Solution 2 tested in France: a positioning algorithm (multi-sensor acquisitions with an emulator of EGNOS DFMC) can result in error values below three metres, in post-processing mode with a favourable GNSS environment. Average distance error was below 0.2 m and average speed error was below 0.02 m/s, but these results were not from a real-time algorithm execution and did not include track selectivity functions.
• Solution 3 tested in Germany: a real-time implementation (dual channel of GNSS, IMU or Inertial Measurement Unit, speed sensors and Digital Map) and used the first track position as a given position (no track discrimination). Due to unfavourable GNSS conditions, this solution’s speed estimates (sometimes above c. 10 m) were generally less accurate than those in solution 2.

Overall, more work still needs to be done on Stand-Alone FSTP, compared to Virtual Balise (VB)-based train positioning. These improvements may involve updating the architecture for Stand-Alone FSTP and looking for more commonalities with (VB)-based train positioning.

Conclusion

Demos of Virtual Balise (VB)-based and Stand-Alone Fail-Safe Train Positioning System highlighted the great potential of GNSS for train positioning, as part of a safe multi-positioning concept.

Calling on highly accurate real-time information from Virtual Balises, supported by other sensors and technologies, this development will enhance rail safety, reliability and efficiency. It will also lead to the phasing out or reduction of the number of traditional train-detection systems, especially physical balises installed on the tracks, which could drive major cost savings for infrastructure managers – as it estimated that the cost of maintenance for a Eurobalise, for instance, is up to €300 per unit.