FAA Files Reveal a Surprising Threat to Airline Safety: US Military's GPS Tests | Hacker News Reader

FAA Files Reveal a Surprising Threat to Airline Safety: US Military's GPS Tests

spectrum.ieee.org

43 pointsblendo47 comments

The FAA has issued a recommendation that pilots who are experiencing a safety problem that they believe to be related to military electronic warfare exercises should use the term "Stop Buzzer." This term, borrowed from military range safety practices, is intended to make it clear among controllers that they should use a procedure specified with the NOTAM to request immediate termination of the exercise or testing. I am not especially clear on the criteria for use of this procedure, I'm not sure if a controller would comply without a declared emergency. I'm also not sure how quick the response will be in all cases - White Sands controllers are in close contact with Albuquerque Center but you'll probably see a slower response if you're talking to someone else or 121.5.

This recommendation came out because there are multiple cases where pilots have declared emergencies related to GPS jamming but controllers have not requested termination of the jamming because of miscommunication or not realizing that the emergency was related.

This is all pretty new and it has not really been communicated very well, I only know about it because of something I saw in _AOPA Pilot_. It has been reported that some aircraft, I forget which but I believe a light jet, react particularly badly to loss of GPS and it can get as bad as loss of attitude information, potentially as a result of a bug.

This is probably the one you're thinking of:

"The Embraer EMB-505 Phenom 300, for instance, entered service in 2009 and has since become the world’s best-selling light jet. In 2016, the FAA warned that if the Phenom 300 encountered an unreliable or unavailable GPS signal, it could enter a Dutch roll (named for a Dutch skating technique), a dangerous combination of wagging and rocking that could cause pilots to lose control. The FAA instructed Phenom 300 owners to avoid all areas of GPS interference."

And this is why it's mandatory to check GPS availability when planning an IFR flight to an airport that only has GPS approach procedures. And it's also required to plan an alternate airport that has a non-GPS approach available.

I’m not too worried about commercial airlines. I’m more worried about press-button-and-fly weekend IFR warriors.

I would also be worried about systems like Cirrus’ VisionJet Safe Return, that turn the aircraft into an autonomous vehicle if a passenger activates the system due to an incapacitated pilot, as GPS is required for its automated path planning to autonomously return to an airport and land.

I haven’t investigated, but I would be curious how such a system responds when GPS is degraded or unavailable.

Surely they also use GLONASS and Galileo.

That’s already a rare event though, so the probability of it coinciding with a GPS outage is smaller still. Time integral of risk exposure and all that.

Rare as it is—and I’d argue it’s not as rare as some people think, having experience three avionics GPS loss-of-signal events in the past two years (in different aircraft)—they’d be nuts not to have considered that risk in the design. I can’t find the AFM supplement for Safe Return but to a first approximation, it wouldn’t be out of the question for it to go into a heading and altitude hold until it reaquires.

It's not required to plan an alternate that has a non-GPS approach if you have a WAAS-enabled GPS receiver, which is almost all of them these days. You can file a primary and alternate with only RNAV approaches.

WAAS won’t help you when the GPS signal is being jammed. It’s ground-based navaids, pilotage, or stay home.

That's a non-sequitur. The person I was replying to said you always have to plan at least one airport with a non-GPS approach.

You’re right of course. I was addressing the larger question of the impact of GPS jamming tests on aviation in general. Less of a reply to you and more of a note to anyone else reading.

But you are still required to check GPS availability and these jamming tests are published way in advance.

I don't remember saying you didn't... Of course you still have to check RAIM availability, but you don't have to plan an alternate with a different approach type.

That doesn't really have anything to do with the GPS jamming tests. That's because GPS isn't actually global, or at least wasn't a couple decades ago. What you're checking is the number of satellites available on your flight path at the time you'll be there such that you can create an accurate and redundant 3d fix.

You’re talking about RAIM. But you’re also required to check NOTAMs, where these tests are documented, under the “all pertinent information” clause of the preflight regs.

You're required to check both, RAIM availability (unless you have a WAAS type GPS) and also the NOTAMs where these jamming tests are published.

On the west coast it has pretty much become a standing Notice to Airmen (NOTAM) that some agency will potentially be interfering with GPS signals. Sadly they continue to decommission VOR stations across the country. https://bruceair.wordpress.com/tag/vor-decomissioning/

This article is really glossing over the fact that non-military jamming now occurs all the time, courtesy of truckers and criminals. Actual GPS jamming is only done rarely on special military ranges with significant separation from civilian traffic, or very rare wide-area trials (days when I would not have flown anywhere near the area in the NOTAM).

One thing is true: GA and Amati pilots rely too heavily in GPS. If they can't fly with DR+VOR in VFR then maybe they shouldn't fly through an area with GPS interference...

Try explaining that to the FAA. They've been shutting down more VORs and airways every year.

The FAA is guilty of cost focused magical thinking. They had some idea that anti-jam techniques would migrate to civilian platforms, but that is delusional, especially as the number of GPS enabled in flight devices has exploded. Maybe for special certified devices for ADS-B. But it is extremely sensitive tech, and anything smarter than a noise jammer can be difficult to deal with.

I think they plan on having jamming surveillance nodes set up near airports. But of course, detecting isn't fixing. How long did that (imaginary?) drone take out Heathrow airport for?

It's not that awful of a problem. Simply making the antenna directional enough is a solution, such as using a phased array coupled with an IMU to tell if the signal is coming from a known satellite location.

It depends on how severe and consistent the jamming is. If the receiver has a good lock (and a good clock) and a basic IMU then jamming on the last few minutes of approach might not cause too much drift -- but it could certainly spook a pilot into diverting.

As to mitigation, it is all expensive (in terms of antenna construction and processing) and GA is price sensitive. Airports with urban buildup get multi-path interference in the GPS signal and jamming signals, and it becomes complicated to computationally null the jamming signals. Chirp jammers are now as cheap as basic noise jammers.

Anyway, it's not so easy, look at any current research paper on anti-jam. Imagine if 500 MAGA/Antifa types plugged in PPDs over a city wide area and kept them mobile? It would be chaos.

Why are truckers causing GPS jamming?

Logging software uses GPS. If they jam it they can go off route or otherwise not have big brother watching them. I imagine it's becoming somewhat less prevalent though, as it's widely known they do it, enforcement has ratcheted up, and it's obvious when trucks just drop off the face of the earth all the time. It's not like 15 years ago when cell networks were much more sparse.

At a guess, I'd say the way they drive is tightly controlled by hiring companies via GPS logs (eg too many / not enough breaks, no driving over / under speed).

Driver fatigue rules often require mandatory rest stops. I. The past this was achieved by writing the rest break in a log book and taking some more uppers. Pesky GPS trackers showed that the break wasn't being done.

It's also traditional to take a load to a yard and rip out the good stuff and reseal it. Chain of custody in shipping is pretty weak. So much more painful if being GPS tracked.

Delivery drivers get rewarded for making all their runs on time, and getting streaks of on time runs, but this is often impossible. So they hand off some deliveries to third parties. But that is very painful if your vehicle is being tracked. GPS is often wonky in urban areas so hard to know for sure if a driver is fuzzing some runs.

As fleet management software gets smarter it will become pointless. But there is a disincentive to fixing log forgery -- cheaper freight. Deaths are externalized costs.

Curious -- how feasible is it for an aircraft to simultaneously support multiple systems for redundancy (e.g. GPS/Galileo/Glonass/BeiDou)?

I don't know how common/feasible it is for aircraft, but most new GPS receivers are capable of receiving two or three of the systems (usually GPS & Galileo or Glonass)

Totally possible. It's also very easy to jam any or all of them with a $50 box.

You'll need a bit more than 50$ to jam them for an aircraft, though

Depends where the plane is.

High altitude, away from any precision navigation areas, not a huge risk.

Flying a GPS approach to an airport - definite risk.

Bear in mind that a GPS satellite has a power budget of ~4kw, and it's >12,000 miles up.

A jammer may only have a watt or two of transmit power, but it can put all of that into a single channel, and it's only going to be a few miles away from your receiver.

You're off by around an order of magnitude:

https://apps.fcc.gov/els/GetAtt.html?id=110032&x=.

Transmitter power is only 25W. EIRP is ~450W.

Of course. But then you'd have to do it while right next to an airport, which is a quick ticket to jail.

In practice, the aircraft will do a go around then proceed with ATC vectoring or inertial navigation.

Ah no. Directional GPS jamming is quite effective with an easily sourced satellite TV dish, a tripod and some straight forward electronics. You can even point it at a plane using their own ADS-B signal.

PPDs are omnidirectional, but quite hard to precisely geolocate when intermittent and moving, as was the case in the well known Newark events (which shut down a major airport).

Airports with difficult approach (like the old Kai Tak in HK) could easily result in CFIT with only small errors, let alone other aircraft.

Big commercial aircraft do that and more (including inertial navigation, older systems like VORs), combining and error correcting all available inputs. Long range air travel existed before GPS was available to civilians, but a certain unfortunate incident showed the need for more precision.

As I understand it, the frequency bands for the various GNSS systems either overlap or are adjacent (see, for example, https://gssc.esa.int/navipedia/index.php/File:GNSS_navigatio...), so interference with one means interference for all of them.

Technically, very easy.

But (IIRC) those other systems are not "officially recognized" by either the FCC (frequency protection) or the FAA. So an airplane operator cannot use them for navigational purposes without opening themselves up to litigation because they're not following FAA directives.

FCC recognition of Galileo is limited in scope:

* https://www.gpsworld.com/fcc-approves-use-of-galileo-for-non...

As a practical matter, jamming of one GNSS would most likely affect the others as well. Particularly since multi-GNSS receivers usually only support the GNSS that share bands (GPS, Galileo, GLONASS)

Perhaps the most compelling option for multi-system redundancy has been the reintroduction of a complete terrestrial navigation system. There have been perennial proposals to fund the construction of an "enhanced LORAN" system to cover the United States that have gotten as far as early design, basically ever since the decommissioning of LORAN-C, but it's never quite happened. Somewhat contrary to what a lot of people might expect, the technology does exist to build a completely earth-based radiolocation system with substantial (but not complete) land coverage, it's just a pretty expensive infrastructure project and has a hard time getting any traction at a time when the FAA is slashing the budget for existing navaid technology.

The infrastructure that was already "half way there," e.g. FAA VOR/DME amd the Coast Guard DGPS infrastructure built on the back of the cancelled GWEN project, is largely all decommissioned or in a slow process of decommissioning.

ASAT warfare has lead to some renewed interest, but not enough to actually get the money on the table.

they should also have multiple redundant IRUs

Redundancy requires an alternate system of navigation information, GPS/Galileo/Glosnass/BeiDou are all satellite based navigation, and tend to use the same frequency range, thus are all subject to the same systemic risk.

True redundancy requires inertial dead reckoning, celestial navigation, or land based navigation aids. Sadly, none of these are being widely deployed.

Their question seemed to refer specifically to satellite navigation system redundancy feasibility.

Is your response "it's not feasible because when one fails, they all do"? Or is your response "it's not feasible because when one fails, they all could"?

There is an important distinction between those two.

Satellite systems fail, there've been quite a few threads about problems with Europe's Galileo system, for example. There can be solar and other astronomic events which cause satellites to become unavailable. The GPS system became available to the public because of political decisions, it could be come unavailable for those reasons as well.

Making the entire country dependent on GPS limits the options of the military organizations that established it in the first place, and compromises their mission to provide navigation to troops in the field. (in my opinion)

The vulnerability of satellites to deliberate attack should not place the entire nation in a position of vulnerability as well.

A ground based system in friendly territory, with a large number of separate sites, provides a redundancy in numbers that you just can't get from a few tens of satellites.

There should also be ongoing research into improving celestial and inertial navigation systems as yet another backup.

A long range system similar to the Omega system should also be deployed.

I've always wondered what GPS degradation (i.e. during tests) would do for things like Required Navigation Performance operations...

Having flown into Queenstown in NZ several times when it's very cloudy, it's quite impressive how they're able to maintain operations apparently without seeing much, when there are mountains very close and depending on the approach, the aircraft is basically flying down a valley. I don't even think that airport has an ILS system for final approach...

https://www.youtube.com/watch?v=Rhkbia1dMcI

> I don't even think that airport has an ILS system for final approach...

NZQN approach plate using VOR/DME:

* http://www.aip.net.nz/pdf/NZQN_43.1_43.2.pdf

* http://www.aip.net.nz/NavWalk.aspx?section=CHARTS&tree=Queen...

It's a 4.9M PDF that loads at about 13 kB/s over ~4 minutes.

Because the valley is curved, ILS cannot be used as a straight-in glide slope is not possible, so you have to 'swing around'.

Maybe I'm missing something, but does El Paso not have localizers/ILS?

GPS isn't used for the final, why did a GPS outage put this flight in danger?

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

There is ILS, but only on RW 22 -- https://flightaware.com/resources/airport/ELP/IAP/ILS+OR+LOC...

Based on the article, I assume they tried that a few times but (cross)winds made it too risky.

Reply on news.ycombinator.com