Episode Details

Republished Episode - I wanted to resurface this episode because positioning is everything, and it's a great overview of how it works.

Right now you are being bathed in radio signals from about 35 satellites, each one 20,000 km away, and every one of them is weaker than the noise floor of the receiver trying to hear it. That is GNSS, and it is the only technology on Earth that can tell you where you are in an absolute sense.

Sandy Kennedy runs applied research for autonomy and positioning at Hexagon. In this conversation she walks through how satellite positioning actually works, why centimetre accuracy is so hard to get, and what the next generation of positioning looks like. We cover low earth orbit constellations, visual positioning, Wi-Fi, 5G and ultra wideband, and the one thing every one of those systems still quietly borrows from GPS.

If you have ever wondered whether visual positioning is going to replace GNSS, or why nobody has just put GPS satellites in low earth orbit already, this one is for you.

Show notes

Sandy Kennedy is the Vice President of Innovation for Autonomy and Positioning at Hexagon. Her group does applied research: finding better ways to do what Hexagon's positioning products already do, and finding new things they could do in the future.

This episode is a tour of positioning from the top down. We start 20,000 km up with the GNSS constellations, look at what a low earth orbit constellation would change, then come back down to Earth for visual positioning, Wi-Fi, 5G, ultra wideband and the private networks used in mines and warehouses. The through line is that none of these technologies replaces the others. They each have environments where they are strong and environments where they fall over, and the real work ahead is making them hand off to each other seamlessly.

What "better positioning" actually means
Sandy's old grad supervisor used to ask "better in which parameter?" Better depends on the job. For positioning it usually comes down to availability (how often you can get a fix), whether that fix is accurate enough to be useful, and then accessibility: can the receiver be made small enough, cheap enough and cool enough to go where it is needed. Power is not just power, it is also heat.

How GNSS works
Every constellation (GPS, Galileo, BeiDou, GLONASS) is a state-owned, state-operated set of synchronised satellites in medium earth orbit. Your receiver measures the time a signal took to arrive, multiplies by the speed of light, and gets a distance. Four satellites give you four unknowns: X, Y, Z and your clock offset from system time. More satellites give you redundancy, and in this case redundancy is a good thing. In open sky today a receiver can see about 35 satellites at once, each broadcasting on around three frequencies.

Why centimetre accuracy is hard
- The signal is below the noise floor of your receiver. There nto. You have to fish it out.
- Broadcast orbits are only accurate to metres. That is remarkable for something 20,000 km away, but if you want centimetres you need precise orbits from
  a correction service.
- The troposphere delays the signal and changes with water vapour. The ionosphere is dispersive and tears code and carrier apart, and it follows the solar
  cycle. Multi-frequency receivers can observe and remove mostally.
- Multipath. In a prairie there is nothing to bounce off. Over water there is more. In a city you are surrounded by hard metal and stone, and the receiver
  has to work out which arrival was the direct line of sight.

Why not just put GNSS satellites in low earth orbit?
Daniel pitches it as a startup idea and Sandy takes it apart, fairly. LEO is cheaper to launch to, satellites need less radiation hardening, the signal
arrives stronger and cuts through foliage better (not buildingte slices through the atmosphere in a way that helps separate