Hacker News Reader

Comment by mapt | Hacker News Reader

Chip-scale atomic clocks based on cesium were demonstrated in 2003 with DARPA/NIST funding, and entered commercial production in 2011.

https://en.wikipedia.org/wiki/Chip-scale_atomic_clock

Apparently they're not even export-protected, despite their obvious use in GPS validation schemes and in RTK.

> The SA.45s CSAC has an Export Commodity Control Number (ECCN) of EAR99. This means it is not ITAR-controlled and does not require a special license to ship to most nations. The SA.45s CSAC classification is controlled by the Bureau of Industrial Security (BIS) within the US Department of Commerce.

The article talks about quantum "optical clocks" but doesn't really explain the concept.

Which appears to be this:

https://en.wikipedia.org/wiki/Quantum_logic_clock

Which, like many things named "Quantum", still doesn't really explain how you get an IMU out of it.

View on HN

The article is deeply confused.

It’s true that optical clocks will improve the accuracy of our measurement of time, and it’s true that GPS depends on time, but there are several steps between primary frequency standards (ie, optical clocks) and GPS, and several more steps between GPS and navigation applications.

So optical clocks cannot, in fact, have any effect on the end-user perceived reliability of GPS.

For that, the best solution is to revive LORAN which is much less susceptible to jamming. (And would also benefit from better atomic clocks.)

Much of Finland and Estonia are currently being jammed per https://gpsjam.org/?lat=58.53948&lon=24.82400&z=4.9&date=202...

Finland is reintroducing DME: https://www.reuters.com/business/aerospace-defense/three-fin...

Which seems to be a different concept from LORAN, but still useful for navigation when multiple base stations are in range.

https://en.wikipedia.org/wiki/Distance_measuring_equipment

You can do about 5x worse (in accuracy terms) than a Cesium clock in a smaller package using a rubidium atomic clock. Average ~4 of these and you get to the same accuracy as a cesium clock. They aren't export controlled because they aren't that special in terms of what you get.

To improve instrumental accuracy by 5x in a single dimension when fighting against random uncorrelated drift/noise, from what I recall of statistics you require 5^2 = 25x as many instruments.

Yeah, I think you're actually right. I was thinking you double them with 2 (experimentally), so you double that again to get 4x, but my curve was pretty off.

They’re probably talking about this for quantum navigation

https://www.theguardian.com/science/article/2024/jun/15/lond...

The diamond-based quantum IMUs are a completely different appliance and a different application (dead reckoning).

High quality dead reckoning over a long duration + an initial fix solves the reliable instantaneous absolute fix. So different technically but OP is correct it would be relevant to the problem of solving GPS jamming.

It absolutely does not. Dead reckoning has error that accumulates over time, and even "high-quality" dead reckoning will be beaten by a crappy GPS fix very quickly.

The dead reckoning I’m referring to (and I suspect op is too and seemingly you are as well), is the work being done by the military in the usage of submarines that stay submerged for extremely long periods of time. The error accumulated is many orders of magnitude less than traditional accelerometers + gyroscopes over the same time frame. The point is you can dead reckon within your error bounds even when GPS is unavailable and the accuracy from fusion will beat GPS by itself (not that that matters for the applications we’re discussing). For the duration of a flight it should be well within the capabilities of such sensors to dead reckon accurately from a last GPS fix before blackout and OP is correct this would be a complementary solution to more accurate clocks making it harder to jam GPS in the first place.

Indeed it’s being examined precisely for this application:

https://newatlas.com/aircraft/quantum-navigation-infleqtion-...

Reply on news.ycombinator.com