Galileo: How Europe Built Its Own Satellite Navigation System

A clock that drifts one second in around 3 million years sits at the core of Galileo, Europe's global navigation satellite system. The passive hydrogen maser flew on the second experimental satellite, GIOVE-B, and the reason such precision matters is simple: navigation is measuring time, and a 1 second error would mean a position error of 300,000 km, enough to put a user on the Moon. The story of how the European Space Agency (ESA), the European Commission and partners built Galileo from nothing, from the 1990s to the present, is told by the engineers who did it.

Why Europe built its own system

In the 1990s Europe watched GPS grow. The system suffered major positioning errors and was affected by selective availability, an intentional degradation of the signal decided by the US government that worsened positions by up to 100 m in Europe. The US government removed selective availability in 2000. Europe had weighed three options: complement GPS, go ahead standalone, or cooperate. The third was set aside because, as one engineer put it, you cooperate only if you have your own system; without one, you just depend. Europe chose its own system, interoperable with GPS.

Two systems were developed: EGNOS, which augments the GPS signal, and the fully autonomous Galileo. ESA served as the technical and design arm, the European Commission as the political one. A tripartite agreement with Eurocontrol added the operational perspective. The ARTES 9 programme, approved in 1995, was described as fundamental: without it, the seeds of Galileo would not have happened. EGNOS paid off in aviation too. Civil aviation authorities and Eurocontrol supplied its requirements, Inmarsat hosted EGNOS payloads on satellites 36,000 km above Earth, and the first EGNOS signal in space came in 2003. A significant part of the EGNOS ground segment technology is common to Galileo.

Building in under three years

Nobody in Europe could call themselves a satellite navigation engineer, and there was no dedicated programme or funding. Galileo also meant a whole constellation, ground and control centres, launchers and operations rather than single satellites. Normal satellite construction takes four to five years; the first satellite had to be designed, developed, qualified, launched and operated in under three, to secure frequency filings. The team applied what is now called NewSpace methods, postponing the full onboard software until after launch, which saved 6 to 9 months.

ESA built two satellites in parallel as risk mitigation, against the loss of one satellite or a launcher failure. During a test on the second, something failed in a small explosion, and that satellite launched only in 2008. The other, GIOVE-A, launched on 28 December 2005. GIOVE stands for Galileo In Orbit Validation Element, and in Italian the name means Jupiter, a nod to Galileo Galilei using Jupiter's moons to compute longitude at sea. On 12 January 2006 the team in the navigation laboratory sent the instruction to switch on the signals. The spectrum analyser showed nothing at first; a few seconds later the spectrum grew. GIOVE-A also flew a rubidium atomic clock, and the team chose two clock technologies because Europe had never flown atomic clocks designed to last 12 or even 15 years.

Ground stations and control centres

Galileo needs worldwide uplink coverage, which led to stations at remote sites, including the North and South Poles and volcanic islands. The European Commission required them to be on European territory. Building the ground station network took around 8 to 10 years. On a first site visit to one island a volcanic eruption occurred, and the shore had grown 100 m by the time the team left; on a later visit there was an earthquake of 6.8. There are two control centres, a prime and a backup, one in Germany and one in Italy.

Operational launches and the first fix

The two experimental satellites flew on Soyuz from Russia, but the operational rollout needed launches from European territory, so Soyuz was brought to French Guiana. The first operational launch there was also the first time ESA used the Guiana Space Centre for a Soyuz launch. The team spent two months there. After launch, confirmation that the solar arrays had deployed and healthy signals brought what one engineer called huge relief. A second launch followed within a year. With all four IOV satellites in orbit, the first position fix was computed in a laboratory using only four Galileo satellites, and success was declared. The infrastructure was entirely European owned and built. The performance computed from those first measurements is confirmed today.

Ten more satellites followed over the next two years, including the first quadruple launch by Ariane in 2016. On 15 December 2016, after 18 satellites were launched, the European Commission declared Galileo Initial Services. Operations passed from ESA to EUSPA. Users, at a couple of hundred at the time, grew to billions.

Services and the next generation

Galileo today has over 30 satellites in orbit. Its services include:

• Open service: free, works alongside GPS.
• High accuracy service: position down to 20 cm.
• Search and rescue: relays a distress beacon's location and can send a message back to the person in trouble.
• Public regulated service: an encrypted signal for government and authorised users, with anti-jamming mechanisms.
• Emergency warning service: works when regular communication networks are down.

In 2024 two launches of four satellites each brought the constellation to full operational capability, after the last launch in 2021 on Soyuz. Six satellites remain to be launched in the next 15 to 18 months, before the first second-generation satellites, which will have fully digital navigation payloads, electric propulsion, a better navigation antenna, inter-satellite links and an advanced atomic clock configuration. They are being developed by two industrial consortia. Galileo's engineers also point to expanding the model to the Moon and Mars.

On EGNOS, Airbus's decision to equip its A350 with the technology helped adoption, and the first automatic landing using it was demonstrated in 2019.