It would be nice if you could send a POST request to open the door if you want some fresh air.

Someone will probably figure out how to send a request to disable auto-pilot or turn off the fasten seat belt sign.

the airline industry is nowhere nearly as stupid as the software industry with things like this.

the communication between plane and wifi/entertainment system, if there is any, is almost certainly one-way. likely, the wifi system providing this info is receiving data from the flight systems and repeating it or transforming it a bit and providing that.

it would not surprise me at all if the flight attendants have to program everything about the flight into the system prior to departure each flight, and there is no communication from the aircraft at all.

"The computer network in the Dreamliner's passenger compartment, designed to give passengers in-flight internet access, is connected to the plane's control, navigation and communication systems, an FAA report reveals." [1]

(I guess there's some kind of firewall, but we know that those are not always perfect)

[1] https://www.wired.com/2008/01/dreamliner-security/

Not a problem as long as you reboot your Dreamliner every once in a while: https://www.theregister.com/2020/04/02/boeing_787_power_cycl...

IIRC the in-flight infotainment systems are entirely separate from the avionics control systems at the data layer. I recall being told that in some cases even the flight status is actually pulled from a 3p api service rather than hooked into the onboard avionics.

There could be some fuckery via shared power or other non-data systems but that’s probably beyond someone sitting in a seat with standard laptop hardware.

Might be one way, but that doesn't mean you can't DDoS it (by accident or otherwise).

If the latter was true, then "wheelWeightState" (and others) would not work. But, they do work.

Not necessarily. All this could be reported up to airline HQ through the satellite link and then the IFE/customer WiFi could be downloading it again to serve locally.

That would prevent any need for direct connection between the systems.

Is that how it works? I doubt it. But it could be done.

The weight-on-wheels switch reports its in-the-air status over ADSB. It errs in the direction of assuming in-the-air, which was implicated in a near miss on the ground in 2020 between two Air Canada flights[1][2]

1: https://www.tsb.gc.ca/eng/rapports-reports/aviation/2020/A20...

2: https://www.youtube.com/watch?v=nj7nG6gJqsU

Is this the thread where the web developers speculate wildly without facts or evidence and pretend to know more about airplanes than the aerospace engineers who built it?

Maybe it would help if Boeing 737's crash wasn't making the case for such incompetence.

I tried to change the flight level by PATCHing altitude, but it seemed to require authentication. Oh well.

This sounds like a good way to meet some upset people with expensive sunglasses shortly after you land.

It's not like that at all. The sunglasses aren't that expensive.

after you land also open to interpretation

I imagine trying this and meet those people while my name is Mohamed :D.

PS:It is sarcasm and actually this is my name.

Maybe you can take risks like that, but I certainly can’t. I don’t think anyone with my name or skin color would be given the benefit of the doubt for even a moment.

Your comment made my day. Eye opening

let me guess, you experience “random selection” events more than the expected amount?

apologies for joking. it must suck.

Oh yes. Reminds me of the XKCD rand() comic!

DELTE

Please stand by, a DHS agent will be with you shortly

You probably can. I suspect airliner software is appallingly insecure.

Relies very strongly on simple airgapping. Can't do anything to it if there's no wires in the direction you want. Can't remotely hack if there's nothing antenna connected that can talk to flight control. It has the luxury of not needing to do the "limited RCE" that is a modern web request

In the article above, in-flight wifi has an API reporting position, altitude, and velocity. That is a feed from avionics, which renders the claim of airgapped systems essentially null.

They could, for the sake of a ridiculous but clear example, have a display hooked up to the avionics and a camera hooked up to a separate computer which reads the values.

There are various ways of connecting systems while physically guaranteeing one way data flow—a fiber optic link with the transmitter removed from one end and the receiver removed from the other is basically a less silly “camera pointed at a display” and used in the real world.

You could argue the exact semantics of “air gapped”, but for the discussion here that’s accomplishing the same thing. The fact that the passenger network has some visibility into the avionics network is not, in and of itself, any indication of an issue.

“Maybe it’s just paint? Could be anything really”, the Antarctic explorer suggested, gazing at his blackened toes.

+1. Instead of saying "airgap" a term I've seen for what you're describing is "data diode".

The plane has a transponder that reports this information to the ground, ATC, other aircraft etc. The infotainment server has a receiver that gets this data. Or in some cases they instead pull it from a ground based service via the internet. The transponder is not able to receive signals, so it is air gapped.

A quick review of published information reveals this claim as false. A typical airliner FMS feeds information to IFE via gateway devices. The integration is intended to be one-way. Airgapped they are not.

Such analytical delusions are the first step on the road to failing to adequately mitigate threats. As practiced by “it can’t happen here” school of fucking up.

Fortunately, it seems far more likely that aircraft system designers do not rely on any such assumption, and practice defence in depth, particularly with regard to integration. There was a good talk at DEFCON 22 by Phil Polstra on the matter.

Documentation please?

The talk above is available online.

Reminds me of an episode of Leverage where they wanted to hack into Congress and change the text of a bill. In the show, it turned out everything was airgapped so they had to send a person to drop off a paper copy of the compromised bill. Hmm, that was also a plot line in Better Call Saul.

You can use jq's string interpolation feature to simplify this:

    $ curl -s https://wifi.delta.com/api/flight-data | jq -r '"https://maps.google.com/?q=\(.latitude),\(.longitude)"'

Thanks! I was trying to figure this out but I didn't have great Internet access (for some reason...) so I just hacked it instead.

In powershell

    Invoke-WebRequest https://wifi.delta.com/api/flight-data | ConvertFrom-Json | %{ "https://maps.google.com/?q=$($_.latitude),$($_.longitude)"

Interesting how they chose to make more general `vehicleId` instead of `planeId` or `tailNumber` or something. I wonder if Delta's fleet includes other things that have matching APIs to this one. I also wonder how much of their internal system structure one could learn from the `flightId` if they knew about other systems. It doesn't look like much beyond a composite key of otherwise knowable data, but still interesting.

I doubt Delta made this. It’s an official the shelf product that can do ships, trains, planes, etc.

Valid point. It makes a lot of sense in that light instead.

But they also have airplane/flight specific identifiers like “flightNumber”, “flightId”, “noseId(?)” and “airSpeed”. Maybe vehicleId is part of a base class or primary key somewhere and that abstraction is leaking.

Hey, a train has airspeed. :)

Heh, true. I deliberately left out altitude because this is HN but you caught me anyway.

Presumably a train's groundSpeed and airSpeed are the same. If they diverge you have bigger problems than a JSON schema.

Is there a variant of this for ships? surfaceSpeed vs seaFloorSpeed?

A train can easily run in a head- or tailwind in the same order of magnitude as its groundspeed.

I've been tempted to mount a small anemometer to my car - by subtracting the groundspeed from the measured airspeed, one can get the wind speed and direction and figure out whether there is a headwind or tailwind and if so how strong. Theoretically this could be used to drive more efficiently, though the extra drag from the anemometer would probably cancel out any gains.

Wow I got HN’ed twice in one thread. This is not my day.

Well, ships have apparent wind vs real wind and apparent course vs real course (currents and drift are a thing).

Those could apply to an air taxi service or something using helicopters as well.

"Official the shelf" - that your new iOS 17 update helping you out? ;)

Not all aircraft are planes.

> "airspeed": null

[nervously looks out window]

That’s just a sampling error. -NaN is when you get scared.

As always, there's a relevant XKCD:

https://xkcd.com/2170/

What makes it so that you can only resolve the host wifi.delta.com during a flight?

I assume the DNS server on the in-flight router is programmed to resolve that hostname to some local device.

Similar to how I can log into my ASUS router from my home wifi by visiting asusrouter.com.

I have nothing insightful to add, I just want to say thanks for posting this!

I’m on a flight right now and just went to this URL. Sure enough, it works!

I know this information is available via the wifi portal’s UI, but a JSON blob just hits different.

```

{"timestamp":"2023-09-28T21:57:39Z","eta":"23:45","flightDuration":164,"flightNumber":"DAL992","latitude":47.4557876586914,"longitude":-111.73490905761719,"noseId":"3883","paState":false,"vehicleId":"N883DN","destination":"KMSP","origin":"KSEA","flightId":"N883DN_SF_20230928195737","airspeed":null,"airTemperature":null,"altitude":35273,"distanceToGo":13,"doorState":"Closed","groundspeed":499,"heading":95,"timeToGo":107,"wheelWeightState":"Off"}

```

Apologies for the JSON formatting, I’m on mobile.

It'd be interested to make a little HTML page that can query the api for each airline that exposes something like this and give you an in-flight display on your laptop.