Insights · Part 2 of the series
Time is the Hidden Dependency
Much of what satellite navigation provides is not position but time, and a clock that is slightly wrong can cause more trouble than one that has stopped.
Ask most people what GNSS does and they will answer navigation and positioning. Underneath, however, a good deal of it is actually a clock.
Mobile networks, broadcast, power systems and financial platforms all need to agree on the time, often to within microseconds. Satellite time is a cheap and convenient way to get it. Cellular network traffic management is all about timing and this used to be exclusively provided by companies such as Chronos providing time signatures from atomic clocks, but things have become more nuanced.
The 2018 Blackett Review describes a heavy UK reliance on satellite-derived time, much of it in places people rarely look.
The failure that worries me is not the doomsday scenario which makes for attention-grabbing headlines. It is something far more mundane.
In January 2016 a ground software fault left GPS satellites broadcasting time that was 13 microseconds out. That is 13 millionths of a second. According to Chronos, some telecoms clients then spent around 12 hours dealing with system errors. Nothing was jammed. The clocks simply said something slightly wrong, and equipment that relied absolutely on them went wrong too.
That is the point. Loss of signal is visible. A plausible but wrong time may not be. Equipment with a good local oscillator can ride out an outage, but it may have no way of knowing its source has drifted.
Government analysis estimates a UK-wide GNSS outage of seven days would cost around £7.6bn, while noting that sectors with good holdover, such as financial trading, would see little loss. Dependency is uneven, and mostly undocumented.
No antenna would have prevented that January 2016 error: the fault was in what the satellites broadcast, not in how it was received. At SkyShield, where we are developing compact antenna technology intended to improve the resilience of GNSS reception, I treat that as a useful reminder. Better reception is one layer of resilience. Knowing what depends on the signal is another.
A practical question for any operator: do you know which of your systems take time from GNSS, and what they do when it is wrong rather than missing?