Testing software

With increasing autonomy and onboard intelligence, the software used to control a mission – either on the satellite in space or in the mission control systems used on ground – has become enormously complex.

Today, a spacecraft like ESA’s Gaia has 42 000 telemetry parameters and 2650 control parameters (comparable to a mid-size jet aircraft) and hundreds of thousands of lines of code onboard. The ground systems that support the spacecraft additionally contain millions of lines of code, spread across the mission control systems, the flight dynamics systems and the ground stations.

It's no wonder that software validation – ensuring that all software components (often supplied by different vendors) work together as they should in a robust ’system of systems’ – is a major challenge.

Further, software is often customised to meet the unique requirements of individual spacecraft performing widely different scientific missions, so each new mission’s ground segment must be validated anew.

“Today, similar missions are grouped into families to exploit some commonalities and achieve some reuse and simplification in the needed software, but we must still conduct extensive validation,” says ESA’s Jean-Christophe Berton, responsible for test facilities and management support at ESOC.

“Manual testing is possible; however, it is a long process and is prone to errors. So the best way forward is to employ automation to conduct testing.”

The GSRF has developed a set of automated tests to speed up and standardise the process. These can run unattended for long periods, over nights and weekends.

This saves time and, more importantly, it enables human experts to focus on the problems and processes that automation can’t handle.

“In the end, though, some aspects of testing can only be done by engineers,” says Jean-Christophe.