Back-ups for GPS: Satellite-navigation systems at risk of jamming | Hacker News Reader

Back-ups for GPS: Satellite-navigation systems at risk of jamming

economist.com

38 pointsjkuria50 comments

I used to work in a government org that tangentially supported the USAF GPS Directorate sometimes. When I left, the military was working on maybe a half-dozen serious GPS "alternatives"/"augmentations"/"backups"/"compliments", and maybe a dozen or so other highly experimental systems to provide similar capabilities.

The issue, as they saw it, was that up until relatively recently, GPS was the only option. Not only did the US military and civilians use it, but foreign forces probably used it as well... maybe even potential adversaries. In a conflict, they wouldn't attempt to disrupt GPS because doing so could actually hurt their own operations. That assumption changed once GLONASS, Galileo, and Beidou (plus the other regional networks) started coming online, all at slightly different frequencies. And, just like the US gives military-grade GPS equipment to its allies, the operators of those systems were probably doing the same, making it more likely that in a conflict someone somewhere will be disrupting GPS because they have another option for themselves.

From a civilian perspective, I'm not all that worried about GPS jamming. Yes, it's disruptive, but most organizations and systems that rely on GPS for navigation or timing (GPS is just clocks in space) should have training or failsafe fallbacks by now, to at least provide a temporary alternative. What does worry me is GPS spoofing. This is why the military has encrypted signals for its own and allied use.

I'm sure the Economist article mentioned all or some of this, but I don't have a subscription.

Edit: Typo fix

> That assumption changed once GLONASS, Galileo, and Baidu

Baidu is the failing Chinese search engine. Beidou is the Chinese version of GPS.

At the risk of a tangent, Baidu is failing?

This is really going off topic now, but yes, I don't see a bright future for Baidu. You have to realize most Chinese people no longer even do searches on the web; they almost completely live in walled gardens like Tencent WeChat (for general communication), Taobao (for commerce), etc. The web is no longer how most Chinese obtain information, news, or do online transactions. So a search engine for the web has become unnecessary.

I use Baidu for a "second opinion" on my searches. Also Yandex. As far as I'm concerned, Google is failing (since I need to do this).

Ha, thank you. Fixed

So... is there a reason that all powers involved wouldn't just use each-others' civilian navigation signals, as the "compliment" for their own military navigation signals? And, as far as they can manage to cryptanalyze them, use each-other's military navigation signals as well?

I.e. why wouldn't you—for any value of "you"—just build your ICBM to be guided by GPS, GLONASS, Galileo, and Beidou (if you can), cross-checking their results against one-another Mars-rover-processor-consensus style?

Why would you rely on something like GPS for something as expensive as an ICBM, when you can have inertial guidance?

Ring laser gyros, for instance.

The FAA should reverse its moves away from ground based air navigation systems, and up the game to include the entire US. The Navy and Coast Guard should get Loran back up and running. Society shouldn't be only one critical path away from collapse. We need backups.

[Edit/Add] I think that the FCC should include ground based navigation beacon transmitting as part of any terrestrial base station licensing. If every AM, FM, TV, Cell Phone station transmitted location/timing information that would greatly increase the number of sources to cross check, and increase coverage. With everything being SDR driven these days, it's just a software update.

The FAA is planning to have complete coverage of the CONUS with VORs with their Minimum Operational Network (MON):

* https://bruceair.wordpress.com/tag/minimum-operational-netwo...

600 VORs will give complete coverage at 5000' AGL.

They are (were?) doing some research around using a network of DME stations. The research was mostly about how to improve the accuracy of TOF measurements by changing the envelope shape to reduce (or at least measure) the effect of multipath.

See:

> The FAA's network of Distance Measuring Equipment (DME) NAVAIDs will provide a PBN-capable† backup to GPS; however, for aircraft without scanning DME receivers (DD) or DD with Inertial Reference Unit aiding (DDI) equipment, the FAA will provide a conventional navigation backup service based on the VOR MON. The VOR MON is designed to enable aircraft, having lost GPS service, to revert to conventional navigation procedures.

* https://www.faa.gov/about/office_org/headquarters_offices/at...

* https://www.federalregister.gov/documents/2016/07/26/2016-17...

† Performance-Based Navigation

The US DOT has been studying this for a few years now, and they concluded in January 2021:

> The demonstration indicates that there are suitable, mature and commercially available technologies to backup or complement the timing services provided by GPS. However, the demonstration also indicates that none of the systems can universally backup the positioning and navigations capabilities provided by GPS and its augmentations. The critical infrastructure positioning and navigation requirements are so varied that function, application, and end-user specific positioning and navigation solutions are needed. This necessitates a diverse universe of positioning and navigation technologies.

* PDF: https://www.transportation.gov/sites/dot.gov/files/2021-01/F...

With modern software, shouldn't it be possible to create accurate kit that can determine the location up to few hundred feet based on pictures we take of the Sun or moon, compass and altitude reading? I thought sailors in the old times could do it with just their head and primitive tools, with modern cameras and compute power I figured we can do better.

In Ben Rich's SkunkWorks, he said the SR-71 had a similar system that would lock onto stars, but I am not sure if it could work when its cloudy, something that's not a problem when you're flying above the clouds, but should still be handy for the airforce or even navy if they can get small drones.

The way I always imagined satellite star sensors was like then-cutting edge megapixel cameras but a lot of early “star tracker” sensors were more like a single intensity sensor that rotates on an axis.

As the sensor(or sometimes the whole satellite) rotates, there’s wide blanking period where sensor is registering the Earth, then there will be series of peaks with specific known range of intensities, like ping ping pause ping ... that ends with sensor registering strobe from the Sun. Either the time from one peak to another or time from end of blanking to some peaks can be measured and “correlated against database”, in reality added and multiplied with just few int values, and positions as well as angular velocity can be calculated.

Being familiar with PC culture, I feel like I tend to overestimate complexity of then-classified military technologies. It’s pleasantly surprising how simple and elegant those classical systems can be.

Fascinating! I'd always wondered how they made it work with 60s-era technology, but when I searched I couldn't find much info on it. This answers some of my questions.

Any pointers on where I can read more?

http://www.prc68.com/I/StellarTime.shtml#P

The Northrop patents are probably the most interesting. IIRC, if you trace the prior art, you’ll find some patents from the early days of WWII.

How they work:

* Telescope on an alt-az mount, preferably on the ground or a gimbaled platform linked to an INS.

* Optical wedge (prism) behind a telescope to nutate the image around the optical axis

* A rotating shutter behind the wedge, centered on the telescope’s optical axis (often a starburst pattern, but several of the patents propose different shutter patterns to mitigate the effect of background luminosity gradient). The shutter is phase locked to the prism.

* Ground glass screen at the telescope’s focal point

* Single-pixel optical sensor (PMT in the 1960’s, but you’d use a photodiode today) sensitive to the entire screen

* lock-in amplifier synced to the shutter (note, however, that the patents do not describe it as such)

The lock-in senses the output from the PMT. Phase and magnitude from the lock-in operate the azimuth and altitude servos of the telescope to center on the tracked star.

When the tracked star is centered in the telescope, it makes a circular pattern on the screen, modulated by the shutter. When the star is off-axis the circular path is offset, so the instantaneous modulation frequency depends on the phase of the prism (it helps to see the figures in the patents).

They also put an IR-pass filter somewhere in the system to cut down on scattered light from the atmosphere, which helps to improve the SNR. I’m not sure the filter is strictly necessary because the LIA should give you dozens of dB of processing gain and navigation stars are visible to the eye in a telescope during daytime—but it seems like “free” SNR.

I couldn't find the picture so spoke from memory and the details were quite wrong, but looks like "canopus sensor"[1] on Mariner spacecrafts was what I had in mind.

1: http://petermasek.50megs.com/starplot.jpg (mirror: https://images.app.goo.gl/9xAn7ATPMSKYLEHs8)

This is why all modern navy ships still carry sextants and train extensively in celestial navigation. It's pretty much a forgone conclusion that GPS will become immediately useless in any large scale nation state war of the future.

Yeah, maybe we can have a "digital" sextant that instead of relying on specialized training to humans, just pushes all the sextant data plus altitude, barometer, universal time (UTC), timezone and sunset/sunrise from a perpetual calendar and such inputs to an android gizmo that gives us an approximate location, something that can be sold commercially for hikers, boaters and everyone for fun.

Automated stellar navigation systems have been around for decades.

https://www.thedrive.com/the-war-zone/17207/sr-71s-r2-d2-cou...

A sextant requires a flat level horizon, and sky clear enough to see the sun, or stars at night, and a compass that indicates true north (with correction for magnetic declination). In an urban environment, there is usually not a clear shot at a flat horizon, nor an area free from metal structures for far enough to get an accurate magnetic north.

Shouldn't a gyroscope compensate or at least allow getting flat level? I think maybe they didn't use it because they didn't have accurate gyroscopes we have now.

For clear shot, would tiny drones that go up work? How high do we need to get a clear shot?

Go outside, do you see a flat level horizon?

You could use the accelerometer to get a good idea of "down" but not enough resolution for navigation.

All the more reasons to use something digital, isn't it? Would make it much easier to figure out the horizon without actually seeing it.

Or you can simply buy an artificial horizon for less than US$ 40:

* https://www.davisinstruments.com/product/artificial-horizon/

* https://www.starpath.com/catalog/accessories/1840h.htm

* https://www.celestaire.com/product/davis-artificial-horizon/

They've been around since (at least) 1907:

* https://www.scientificamerican.com/article/an-artificial-hor...

You cannot escape the burden of training people to do it, because someone has to check the machines figures for drift due to defect, improper implementation, or damage.

See the Patriot missile systen fiasco.

Navy also uses inertial navigation. It needs to be calibrated with GPS every now and then but you can get away without a GPS downlink for a while before drift becomes a problem.

I bet it's also why cruise missiles use terrain navigation.

Yup. This open source software will do that for arbitrary star fields: http://astrometry.net/

You still need the camera’s orientation relative to earth and time to get your coordinates though.

> With modern software, shouldn't it be possible to create accurate kit that can determine the location up to few hundred feet based on pictures we take of the Sun or moon, compass and altitude reading?

Unless it's cloudy.

Don’t you also need a clock to do this?

20 years ago a similar venture in the UK and Australia attempted to create a dedicated network of towers for locating stolen vehicles. It didn't survive. I think the beauty of GNSS at this point is that if you muck with it the whole population will notice, and that is kind of "bad media". Not exactly winning "hearts and minds".

While IIRC GLONASS, Beidu and Galileo have less than full-planet coverage, https://novatel.com/support/known-solutions/gnss-frequencies... suggest there are a reasonable range of frequencies in use. As I am not a radio guy, in terms of countermeasures, then, would using multiple GNSS systems in places where their constellations are visible frustrate jamming, since standard ICs now support all systems?

In cars and boats, inertial navigation is also common. This is what provides the "you are x% through the tunnel" movement on car navigation screens. In the event that GNSS is lost, just like driving through a tunnel, one would expect operation to continue at a lower resolution / slowly increasing error on many of these systems.

In addition, there are alternate systems for shipping. I believe AIS publishes last known GPS position/heading, which RADAR and depth-sounding can confirm (along with terrain knowledge). By utilizing the shared state of nearby vessels prior to and during an outage, presumably inertial navigation assumption related error could be greatly reduced (as per the DGPS idea that IIRC was common prior to the unmasking of higher resolution GPS signals earlier in its public-use evolution).

Finally, free data sources for celestial navigation systems, terrain-based navigation systems (including bathymetry) and cell sites are available which can provide pretty reasonable estimates as well.

I think it's a safe bet that, even more so than at present, future systems will typically operate on "sensor-fusion" to de-risk outages in sensitive applications.

>As I am not a radio guy, in terms of countermeasures, then, would using multiple GNSS systems in places where their constellations are visible frustrate jamming, since standard ICs now support all systems?

So rather than buying one jammer, the bad guys will just buy several jammers?

How can you do inertial navigation for a car inside a tunnel? afaik the only info you have is the speed and distance covered, from the wheels, but that info can't be shared unless you hook up with the car through obd2 or maybe android auto, the phone itself can't do it, right?

In addition to the other answers here, I'd imagine many stock navigation systems are hooked in to the car's CAN bus which would have speed and sometimes steering angle. Obviously getting that data to a phone is a different story.

The car also most likely has a built-in compass.

You can't, mems gyroscopes drift too quickly to be useful.

Most tunnels where GPS "works" pass a repeated GPS signal through a leaky feeder, the receiver uses doppler to figure out your speed and guestimate distance. I guess because of the point source on the surface the phone must also think the tunnel is a vertical shaft as well.

The new ones create take ephemeris and almanac data and create simulated time signals for the repeaters exact location. This is the same thing that military jammers do near important locations like embassies and bases to ensure that enemy missiles hit somewhere across the road.

Inertial navigation means calculating your next location based upon your speed and trajectory. Cars can calculate how fast they are going and when they turn from wheel size, revolutions and steering inputs, they don't necessarily need IMUs to imply this data. Sibling responses in this thread miss this point.

you have gyroscope and accelerometer data which can tell you distance and direction changes up to the sensitivity of the instruments.

Just to add to this, submarines and ICBMs navigate by inertial systems. The good ones (incredibly expensive) are extremely precise. The problem with INS though is that every minuscule error compounds over time... since there is no external reference. The longer you navigate by INS, the less certain you can be. With GPS, the system is constantly getting new reference signals from the satellites (which periodically get their own from ground stations).

> While IIRC GLONASS, Beidu and Galileo have less than full-planet coverage [...]

Full coverage either (a) already exists for all of these systems, or (b) is planned for all of these systems.

> suggest there are a reasonable range of frequencies in use.

Not really. The frequencies bands reserved for GNSS are between 1164 and 1300 MHz, and 1559 and 1610 MHz:

* https://www.orolia.com/resources/blog/lisa-perdue/2019/gnss-...

However there are specific frequencies that things are focused on. If you actually look at where each service has each of its frequencies, especially visually, they're all quite close together:

* https://www.tallysman.com/gnss-constellations-radio-frequenc...

* https://gssc.esa.int/navipedia/index.php/GNSS_signal

L1/E1/B1 are all stacked on top of each other, with G1 being 25 MHz of to their side; L2 and G2 are on their own; L5/E5A/B2A are all stacked as are G3/E5b/B2; and so are L7/E6 with B3 overlapping a whole bunch.

Also remember that a lot of the signals are at quite a low power, so localized jamming is not that difficult.

None of your suggestions are robust enough for the stated needs, especially when it comes to timing. A reliable source of UTC is needed in a whole bunch of things (e.g., electrical grids), and "sensor-fusion" will not cut it. It will not even cut if for navigation.

The US DOT released a study that what will probably be needed is multiple systems at various frequencies:

> In a long-anticipated report issued last week, the Department of Transportation (DOT) outlined the results of its GPS Backup Technology Demonstration project.[1] As officials had previously projected, it called for a system-of-systems approach using multiple complementary technologies.

* https://rntfnd.org/2021/01/20/dot-report-l-band-uhf-lf-and-f...

* PDF: https://www.transportation.gov/sites/dot.gov/files/2021-01/F...

At the very least they probably have to resurrect (e)Loran, or something less in the low- to medium-frequency (<2MHz) range.

None of your suggestions are robust enough for the stated needs

Disagree. Your sole suggestion (resurrect (e)Loran) is expensive and unlikely. IMHO we are more likely to add better geolocation query features to cellular networks than build another network of ground stations.

One I forgot to mention is internet-based wifi SSID geolocation lookups... these are generally super accurate and fast.

> Your sole suggestion (resurrect (e)Loran) is expensive and unlikely.

I said "very least". It may not be the solution, it is probably part of a solution(s). The various options being looked at are listed, with test results, in the DOT document.

And running the Loran-c was not that expensive:

> DHS also reported that terminating the system would save the government $36 million in fiscal 2010 and $190 million during a five-year period, according to its budget.

* https://www.nextgov.com/cio-briefing/2009/03/obama-proposes-...

For the purchase price of two F-35s the Loran-C infrastructure could (have) be run for five years.

(e)Loran is being (has been?) deployed by South Korea because of all sorts jamming around their peninsula, and it's quite robust. Russia never shut down their (Chayka) network. China is expanding their network:

* https://www.gpsworld.com/china-expanding-loran-as-gnss-backu...

> One I forgot to mention is internet-based wifi SSID geolocation lookups

Would Wifi SSIDs will be available to the aircraft flying at 50,000 feet? Which SSIDs are available to folks in their boats in the Florida Keys? Or boats in the shipping lanes in the middle of the Atlantic? Loran-C skywave propagation (especially at night) can give rough positioning at great distances. You wouldn't want to use it for an ILS approach of course.

You apparently confuse operations costs with costs of establishment including site acquisition. Anyway, the die is cast. https://en.wikipedia.org/wiki/Loran-C#Loran-C_in_the_21st_ce...

Both sides in the Donbass war regularly capture each other drones by spoofing/jamming GPS (1. both sides are naturally using non-military GPS; 2. because of the ceasefire the drones is one of the major modes of the war there currently)

Iran also captured intact a big US drone claiming by doing GPS spoofing

https://www.youtube.com/watch?v=g-jdL7FyYDA

I know the usa is working on optical and electronic technology to use the stars even during the day. They can combine that with inertial guidance on cloudy days .

Yeah, because SR-71 was very fast, they had to build special computerized star-based navigation system for it, as this was the early 1960s, a decade before GPS.

I wonder how similar the new and old systems are. All the principal astronomical science hasn't changed much at all, we just got faster compute and better optics.

https://timeandnavigation.si.edu/multimedia-asset/nortronics...

the star-based navigation was common on ICBM back then and i think still is.

Reply on news.ycombinator.com