
By Miguel Munoz – Global rail director – Expleo Group
As signalling has become more software- and data-driven, the number of interfaces engineers have to account for has grown significantly.
An interlocking, for example, no longer sits neatly on its own within a self-contained, hardware-centric signalling environment. It must work alongside rolling stock, track infrastructure, telecoms, cybersecurity, and a growing ecosystem of digital systems spanning different suppliers and, often, different generations of technology.
Today, the next technology leap is underway as rail moves towards increasingly software-driven and cloud-ready infrastructure. This makes integration considerably more complex. Not only must engineers introduce more sophisticated safety systems to increasingly complex rail networks, but they also must ensure each component in that network behaves securely and works as part of a cohesive system.
Take rolling stock. You can’t reliably validate the behaviour of rail infrastructure independently of the trains using it. Engineers need to understand how the train, signalling, and communications work together, and validate their interoperability as an integrated system in a specific operating environment.
That’s often harder than it sounds, particularly as access to rail lines and fleet is limited. Still, the less of that combined behaviour you can validate early, the more risk you carry into the point where train, track, and software meet.

How virtualisation saves engineering hours
Simulation offers a way around some of that dependency on physical access. Rather than waiting until everything can be tested together on-site, engineers can build a digital representation of the train, the signalling infrastructure, and the communications environment, and test how they might interface in a live environment. This brings a critical part of the validation process much further forward in development.
This is particularly valuable in brownfield projects, where integration is often hardest.
Suppose an operator wants to replace an interlocking on a brownfield line with one from another OEM. The new implementation needs to work with the existing infrastructure as well as any control logic, traffic management, and communications systems already in place.
By virtualising the environment first, signalling engineers can understand if and how each independent component will interact before they commit valuable engineering hours to building the interface.
Managing integration across multiple stakeholders can be equally difficult.
The OEM developing the new technology, the supplier behind the existing system and the infrastructure owner all have a role in making the interface work, but may not work to the same requirements or move at the same pace.
Standardisation is only part of the challenge, however. Capability matters too. OEMs tend to have deeper expertise in emerging technologies, while rail operators and infrastructure owners, particularly large public or incumbent organisations, often find it harder to build the same specialist capability internally.
Public procurement can make this gap harder to close when competitive rules limit how closely an operator can engage with an OEM before going to market. This makes it harder to informally draw on specialist expertise while integration requirements are being drawn up.
Addressing integration from the outset
Late integration is expensive. Discovering an interface problem during site testing can mean revisiting configuration and validation work that has already eaten up significant engineering hours.
A better approach is to make integration something we design for from the outset and continuously validate throughout the engineering lifecycle. That means treating interface behaviour as a key component of the architecture itself.
Standards such as EULYNX, which is designed around modular signalling and common interfaces, are a promising step in this direction, because they give suppliers a common specification for how different signalling subsystems connect and exchange information. Digital twins, model-based design, and AI are also becoming critical parts of the engineering toolbox, allowing teams to model interfaces, automate repeatable work and reuse engineering data.
We’ve spent decades industrialising the technology that runs the railway. The next step is to industrialise the engineering around it, creating reusable, standardised, and continuously integrated engineering processes and valuable data from day one.
Ready to rethink your approach to signalling engineering?
Discover how Expleo’s rail engineering services can support your move towards software-defined, integrated systems.





