Using dead reckoning in railway positioning

Modern rail networks depend on precise, continuous positioning to operate safely and efficiently. Yet tunnels, covered stations, and dense urban environments make reliable train positioning a complex challenge. This page shows you how dead reckoning (DR) technology from u-blox technology can help solve the challenge.

What is dead reckoning?

Dead reckoning is the use of an inertial measurement unit (IMU), which contains a range of accelerometers and gyroscopes, run through an algorithm called a Kalman Filter, to estimate a vehicle’s position, heading, orientation and speed. At a systems level, dead reckoning doesn’t rely on GNSS signal to work, and as such works to complement weak GNSS signal, or compensate for no GNSS signal over short distances. At a product level, most dead reckoning systems are integrated with GNSS. The GNSS helps to initialize and calibrate the dead reckoning system, and the two technologies complement each other during operation.

u-blox has developed several DR solutions over the years, including untethered dead reckoning (where DR is used to fill in gaps in GNSS signal) and automotive dead reckoning (DR combined with GNSS signal and data from wheel tick sensors). ADR for rail is the newest DR offering from u-blox, combining ADR with a unique dynamic model optimized for rail.

^{GPS receiver using wheel-tick and drive-direction signals} 

Why does rail positioning need dead reckoning?

Rail infrastructure – often decades or even centuries old – includes long tunnels, steel canopies, and deep urban corridors that block or reflect GNSS signals. For operators, this creates significant blind spots in their ability to track trains in real time, if they rely solely on GNSS-based solutions.

Without reliable position awareness, scheduling becomes less efficient, depot operations require manual oversight, and passenger information systems risk delays and inaccuracies. Traditional systems that rely on intermittent updates or fixed track circuits cannot deliver the continuous positioning required for modern, data-driven railway management.

The importance of the rail dynamic model

The dynamic model in a DR solution contains rules about the vehicle’s motion that help the Kalman filter discard erroneous data. For instance, the model for a car says that data indicating the vehicle has moved sideways with no forward or backwards motion cannot be true, because cars cannot move laterally. For DR in rail positioning, a similar model is needed.

The challenge is that trains are unlike any other vehicle in terms of movement. They are slow to accelerate, but reach very high speeds. Their turning circle is very limited and governed by tracks, so it’s not possible to drive one in a figure of eight (a common requirement for initializing DR systems). And because they have two bogies that move instead of one fixed and one moving axle, there is no difference between forwards and backwards movement.

Capturing these things in a dynamic model has long been a challenge. u-blox ADR for rail navigation has a successfully designed and tested model that ensures the DR solution functions correctly in a rail environment.

The u-blox railway navigation solution

The ZED-F9R is u-blox’s railway dead reckoning-capable hardware module. Railway dead reckoning support got introduced with the HPS1.40 firmware which is pre-installed on the ZED-F9R-04B product version. We chose to integrate railway ADR into the ZED-F9R as it is a mature product with a long expected lifecycle, meaning that rail operators can use it for years, if not decades to come without having to worry about re-engineering their systems.

The ZED-F9R is also capable of standard precision GNSS, making it a completely flexible solution for railway positioning. Together, these technologies create a seamless ecosystem that ensures continuous, high-accuracy positioning – the foundation of efficient, connected rail operations.

Real-world results: Regionalverkehr Bern - Solothurn

^{This RBS regional train was used for the tests}

After automatic calibration within a few hundred meters, the ZED-F9R travelled a total of 2.5 km using just DR, through the Schanzen tunnel in Switzerland to the underground station of Bern and back. The total position divergence at the end of the test? Less than 30 m.

Click below to read more about how the test went, and how it demonstrates the possibilities for future train positioning solutions that function in complex environments.

Read more

Shaping the future of rail positioning

As automation and digitalization accelerate across the rail industry, the need for dependable positioning will only increase. Railway ADR supports GNSS-based navigation across the railways to provide a foundation for a smarter, safer, and more connected rail future.