FMS vulnerabilities highlighed at Net Security conference
 

This article is obviously going for the shock factor (I tried to tone down the headline) but it seems like this guy has found some interesting vulnerabilities?

It's still curious to me how he thinks he can "hack"an FMS via ACARS or ADS-B... I sincerely hope that's hyperbole.

I'm more curious about why people would attend such a presentation. Rationality has departed; credulity has no bounds in our wiki age and this article reflects the current depth and quality of investigative journalism. Frightening people has become an easy pastime.

There is no link between the described systems that will move the flight controls of the aircraft and that leaves the crew at the mercy of an Android phone.

He's talking about a "payload" (think "virus") that could be deployed on the FMS computer. I've no idea how (or if) this can be done, but if it could then it should be relatively easy to modify the FMS's inputs so as to deliver the desired outputs.

While ACARS/ADSB normally have no connection with the flight controls, my understanding is that this connection is precisely what this "payload" provides. But it doesn't have to - it may just contain the instruction to fly into the ground on the 4th of July, for example. Of course, that wouldn't be as cool as controlling the plane through your Android phone.

The real challenge isn't in programming the payload, it's in delivering it (or preventing its delivery, depending on your point of view). One hopes that the security measures involved with program updates (and possibly also nav data) are up to the task.

It's been done with industrial control systems (stuxnet), no fundamental reason why it can't be done with an FMC.

Researcher hacks aircraft controls with Android smartphone
 

Researcher hacks aircraft controls with Android smartphone ? The Register

A presentation at the Hack In The Box security summit in Amsterdam has demonstrated that it's possible to take control of aircraft flight systems and communications using an Android smartphone and some specialized attack code.

Hugo Teso, a security researcher at N.Runs and a commercial airline pilot, spent three years developing the code, buying second-hand commercial flight system software and hardware online and finding vulnerabilities within it. His presentation will cause a few sleepless nights among those with an interest in aircraft security.

Teso's attack code, dubbed SIMON, along with an Android app called PlaneSploit, can take full control of flight systems and the pilot's displays. The hacked aircraft could even be controlled using a smartphone's accelerometer to vary its course and speed by moving the handset about.

"You can use this system to modify approximately everything related to the navigation of the plane," Teso told Forbes. "That includes a lot of nasty things."

First, Teso looked at the Automatic Dependent Surveillance-Broadcast (ADS-B) system that updates ground controllers on an aircraft's position over a 1Mb/s data link. This has no security at all, he found, and could be used to passively eavesdrop on an aircraft's communications and also actively interrupt broadcasts or feed in misinformation.

Also vulnerable is the Aircraft Communications Addressing and Reporting System (ACARS), the communication relay used between pilots and ground controllers. Using a Samsung Galaxy handset, he demonstrated how to use ACARS to redirect an aircraft's navigation systems to different map coordinates.

"ACARS has no security at all. The airplane has no means to know if the messages it receives are valid or not," he said. "So they accept them and you can use them to upload data to the airplane that triggers these vulnerabilities. And then it's game over."

Teso was also able to use flaws in ACARS to insert code into a virtual aircraft's Flight Management System. By running the code between the aircraft's computer unit and the pilot's display he was able to take control of what the aircrew would be seeing in the cockpit and change the direction, altitude, and speed of the compromised craft.

He admitted that some of this was moot, given that the human pilot could always override the automatic systems, but the software could be used to make cockpit displays go haywire or control other functions, like deploying oxygen masks or lights.

The precise nature of the code flaws wasn't released – for understandable reasons – but Teso says the Federal Aviation Administration and the European Aviation Safety Administration have both been informed and are working on fixing the issue. ®



Skipping the usual press over hype - but still.... Really?

There's more information here and his presentation is here (in PDF format).

So with the right radio / software, could you manipulate the ADS-B information that is part of Transponder Mode-S data to initiate a TCAS RA?

I suppose the question is - are there any vulnerabilities in the FMS which allow the FMS to be programmed via ACARS?
If the answer is no, then the attack potential is to retrieve lots of data and send bogus messages and flight plans into it. I would be very surprised if you can send something to an FMS and for that thing to be automatically used / executed without Pilot involvement?

I just read this over the The Register and I suspect that everything he says is possible is indeed possible. These systems were designed with the assumption that both the transmitting device and the receiving device were validated. I'm sure a great deal of time and effort went into validation and testing to make sure the transmitted messages were properly formatted, transmitted, and received. I'm sure the system was also tested for its ability to detect and reject messages corrupted by random interference.

But the complete lack of any authentication security tells me that there was no attempt to validate the system for deliberately constructed malicious messages. In networking systems, maliciously constructed messages/packets are probably the most common attack vector. And they often succeed, even on networks hardened against such attacks. I should think that do what he claims would be child's play for someone with in-depth knowledge of those systems.

But the ACARS network is more like your home LAN than the Internet; there are few legitimate routes into the system and they're validated as trustworthy before they're allowed to send data. An evil person at an ATC centre could send evil messages, but you pretty much have to trust them regardless of how the messages are transferred.

For this to work, they'd presumably need a suicidal passenger on the aircraft carrying a radio transmitter powerful enough to override the ground transmitters. Which doesn't seem too easy to me.

I will look at the feed, but unfortunately, I would assume it is valid.
ADSB issues are one of the biggest reasons that ADSB-IN is not moving forward.

In regards to the FMS, those of us who work with the system architecture, understand the potential vulnerabilities. It is somewhat unfortunate that this issue has been brought forward in a hacker format...

Edit: In looking through the presentation, I dont have the verbiage of how or what was explained, but the presentation is very accurate. :eek:

Well, before we all set our hair on fire over some half-baked notions we need to think about this, with some understanding of the systems involved and not just ride off in all directions with ill-considered claims. The sky is not falling . . .

1. Neither the ADS nor the ACARS are directly linked / connected to the flight control system, period.

2. The FMS is connected to the flight control system when the autoflight / autothrust systems are being guided by the flight plan data and (to a much lesser extent) the weight-and-balance data entered during the ramp check.

2. The ADS system is an ATC communications system which has no connection to the FMS. ATC cannot control the routing, speed or altitude of the aircraft through ADS.

3. While some operators routinely upload flight plan and weight-and-balance data via ACARS during the ramp flight preparation sequence, many operators' do not have this auto-upload capability and the data is entered manually. In manually-entered circumstances there is no way to upload changes to the flight plan routing via the ACARS to affect aircraft navigation through the FMS which is connected to the autoflight system.

4. Given system and aircraft design, logically the autoflight system must be engaged for this to "work". The FMS has the route data and the autoflight is designed to follow that data.

5. FMS data cannot control altitude and will not command the aircraft to climb or descend even if cruise altitude changes and descent points have previously been entered or otherwise programmed in the FMS. Neither can ACARS nor ADS do this.

6. Within a narrow Mach or CAS range, when routinely engaged, the autoflight / autothrust systems are controlled by the FMS which in turn will control aircraft cruise speed. Cruise speed and speed restrictions at certain waypoints, (oceanic entry and exit points, for example), may be part of the flight plan. As with any FMS entries, there are reasonableness checks which reject incorrect or inappropriate data.

7. Should something like that which is claimed actually succeed, there are at least two human pilots in the cockpit, sometimes three or four depending upon phase of flight, etc who can fly the aircraft manually. When the autoflight system is disconnected none of this works. Also, routine enroute waypoint checks confirm position, speed, altitude, next position and so on and, should immediate but subtle anomalies occur enroute, they would be caught at such waypoints.

This doesn't deny the possibility that ACARS has vulnerabilities, but such potential is not about to take over an airliner in flight as implied by the use of the word, "hijack".

In my view, making claims that it is somehow possible to command an airliner to "dive" or do other untoward maneuvers beyond the crew's ability to counter, using an Android cellphone, is irresponsible.

When the exact method by which the claims in Mr. Teso's article are made is clearly explained and, as such things normally are required to be, peer-reviewed to substantiate serious claims of compromise, then we might take all this seriously. At present, it seems entertainment is where one finds it.

PJ2

I would put this topic along with chemical contrails.

FB

PJ2, FB...

I can say this, that after 911, there was a very serious effort in these regards.

With ADSB, there has been ADSB IN capability on many aircraft for quite some time now.

If the aircraft is on a coded procedure, where do the speed and altitude commands originate?

(thats enough at this point, I dont want the black SUV's showing up)

The actual presentation deck from the conference is here if you want to read the actual presentation itself instead of relying on second-hand news articles.

http://conference.hitb.org/hitbsecconf2013ams/materials/D1T1%20-%20Hugo%20Teso%20-%20Aircraft%20Hacking%20-%20Practical%20Aero%20Series.pdf

F.OBN;

Thanks for your response.

I realize that the issue has a security element to it but the point needs to be emphasized that ACARS & ADS systems cannot take over the flight controls of an aircraft, and that isn't a security issue, that is a design feature, knowledge of which is available to anyone. For heaven's sake, no "black SUVs" are going to show up for discussing such an issue!

I'm not disputing claims of interference through vulnerabilities, I am disputing the claim that such vulnerabilities represent a threat to the physical control of airliners beyond the ability of flight crews to counter. Let us not conflate the issue such that all manner of rumour be taken at face value for fact.

In general, let us not raise and embrace the possibility, then refuse to discuss it out of some concern for secrecy or security. If the threat is real, demonstrate how it is thus using commonly available knowledge and information. I have made some points regarding why I think this is nonsense but claim no expertise in any area other than flying these cable-pulley, hydraulic and fbw transports. Tell me as a pilot why such concerns should be taken seriously when there is a flight crew on board that can manually fly the airplane.

When someone here who both embraces these claims (that airliners can be "taken over" by ACARS or ADS commands through the FMS directly controlling the flight and engine controls, autoflight on or off), and can describe the method or process by which this is made possible and cannot be defeated by the cockpit crew, then perhaps we can take this threat seriously.

I would think that the risks are far higher in terms of corporate espionage for data that airlines are always desperate to gather on their competitors. But that kind of hacking is not new and it doesn't threaten flight safety.

nombody, thanks for the link. The preso reminds me of something Von Daniken or Velikovsky would put out.

PJ2

This reminds me of the hype with computers leading up to the year 2000. People made a lot of money claiming disaster was around the corner and carrying out expensive audits. It is (just) conceivable that you could screw up the flight management system. But then what? after all that is what the pilots are there for....to fly the plane. If the system fails we revert to?....manual flight....big deal!

FYI

There is a lively thread over at Hacker News. Several of the developers claim to have experience with avionics software programming.

https://news.ycombinator.com/item?id=5531679

p.s. also a humorous sub-thread about possible TSA responses: ban Android Phones; seal them in one quart ziplock baggie; wrap them aluminum; put Hugo Teso on the no-fly list; etc..

PJ2,

Understand your response.

If this was possible, would post how to do it in an online public forum?

Think about your autopilot controls, there is a button on the yoke, and several other ways to control the autopilot. Can the autopilot disengage itself?

I read the presentation and I think this is a serious problem.

What the security researcher is talking about is using unsecured communication channels (ADS, ACARS) to identify then attack an aircraft, compromising the FMS and possibly other systems.

From what he was saying it appears that there are ''zero-day exploits'' available for some FMCs, through normal data channels. Once in there, the attacker could do pretty much anything. :eek:

We tend to think of aircraft nav/data systems as being made up of isolated units but if there are communications between them, then they are vulnerable. You can do a fair bit with most FMCs: tune navaids, select navigation sources and even use them as backup dials and switches for when these fail. On the 777 you can be ''pushed'' route updates by ATC (or whoever is pretending to be ATC...) Once compromised, you could display to the pilot(s) ''situation normal'' but in fact be taking the aircraft off-route...

No one thought that it was possible to remotely make 30 non-externally networked ultra high speed centrifuges located inside a super secret, hardened nuclear processing facility operated by country no one can get into quietly spin their bearings to destruction.

But it happened.

I'm not concerned about a flight becoming someone's jumbo size Parrot AR Drone, but I would be concerned that system interruptions or system manipulation could be used to provoke a mishap.

All this, and TESO was not an avionics expert...

I think you know you are talking chalk and cheese here !

You know very well that particular network was most likely targeted by well funded, well connected entities possibly linked to one or two governments with a vested interest.

I think a well placed, high power ground transmitter could easily overcome a more distant "legitimate" transmitters.

Even though this discussion has been focused on the ground/air links, I would be willing to wager that the back-end ground link portions of these systems are almost as vulnerable. Why spend money on a transmitter when you don't have to.

Never underestimate the power of money or determination. :)

F.OBN;

Again, thank you for engaging the question. I think discussion of this matter is important so that the salient issues can be discerned against a substantial background noise.

The salient issue is, as stated, that there are certain vulnerabilities to ACARS, ADS, (and also to GPS/SatCom, but that too, is old news; we've known about Satellite / GPS jammers for years). To me, the widespread (and growing) use of iPads for critical flight planning, weight-balance, enroute and approach-plate work represent a far, far greater security risk in terms of hackability with clear and safety-related results, than the present matter under discussion, but we'll stay on-topic.

The background noise is the inability of those proposing the notion that this is a serious problem, to articulate the means by which this is more than just an ACARS / ADS vulnerability issue which is, like the issues mentioned above, probably well known.

The presentation works because it implies much but says very little. We'll see where this all is in a week or so.

PJ2

How did this guy supposedly get around the requirements to ACCEPT, LOAD, and EXECUTE a flight plan change; and override the altitude locked in via the MCP?

Sounds like a lot of sensationalism, or some ACARS and ADS/CPDLC implementations are a LOT different than the ones on the 744 and 748...

PJ2,

Thanks for the response. As you have noted, it is a very complicated issue. It isnt about GPS jammers, or high powered signals from the ground, these are at best, an annoyance.
This about getting to some of the core processes.
Like a thief, you check all of the ways into the building, and the first thing you is when you get to the 32 floor, you lock down the elevator behind you.

How are these commands entered?

(in the 744, there are, from what I remember, 4 places where you can command disengage to the autopilot system, and the system can disengage itself?)

edit: this is just f'n great...this made Yahoo news... How a single Android smartphone can crash an airplane)

I think you and PJ2 may be confusing what the FMC/FMS allows the flight-crew to do (often a company decision limiting training requirements) and what the FMC/FMS software can actually do. Just because they have not put the menu items, buttons and switches on the outside doesn't mean that the software capabilities are not there. Indeed, I would be surprised if there were not several undocumented capabilities put into the software and passed through certification, so that they could be 'delivered' rapidly as upgrades just by allowing the function.
Like many control systems (such as power control systems, communication switchgear, flood gates etc) the firmware and software may have been written without any attempt at defensive coding. There are often hidden codes left over from testing that due to certification issues are not taken out - as nobody would send that code over ACARS would they? Well seems that it is possible. I would be more interested in how he accessed the ACARS frequencies and spoofed the log in from a standard Galaxy phone. But once into ACARS I have no doubt this is possible.

I've seen many hacks that simply use amount and repetition to cause a failure in the code somewhere. The basics are:
1. Analyze code in the target computer.
2. Find some code that will break when presented with a contrived data input.
Note this input can be huge and even presented over time with careful timing.
3. Imbed code in your input that does what you want.
4. Hit the target with this input.
5. When the code breaks the computer blithly keeps on running - right into your code.
6. Computer is now running your code with whatever permissions the broken code had or possibly more or even maybe ALL permission (depends on design, cpu etc etc)

Anyway, the quick version. Years ago I remember being amazed at how the memory allocation from repeated calls was tricked into providing unzeroed memory to the process. Guess what? The process assumed the memory would have all zeros. Bham!

One of the basic problems with code written these days is variable range checking. It takes coding time to write the code, CPU time to execute it, and engineering time to validate it. As code becomes increasingly complex, the penalties expand exponentially, so it is often skipped. This, in combination with "vestigial" code, or deliberately added "undocumented function calls" create an enormous opportunity for exploitation.

It looks to me like these systems were designed under the security through obscurity mantra (after all, who would mess with our little corner of the world?). This, of course, never works, especially today's connected world where nothing is "obscure".

I read the following list of "features" that were demonstrated to work against the simulators by Teso's Android AppSeems like the real deal to me. The paranoid in me would speculate that the powers that be have known about this vuln for a while and this is, in part, one of the reasons for the "no electrics" ban on takeoff/landing (the most vulnerable part of any flight). I can only hope things are patched soon to make tampering more difficult. A hardened fix may require a complete change in architecture.

Areobat
Not sure that would be necessary. All that would be required is a communications gate keeper in firmware that security checks the ACARS/ADS or any other inputs in the same way virus scanners work today. The speed of the scan compared to the lethargic ACARS or ADS would mean that it would not slow anything down.

All these issues will be revisited as the aviation world is dragged kicking and screaming into modern communications systems with Aircraft Access to SWIM (System Wide Information Management). Perversely, in the SWIM world things will be far more secure mainly because it is perceived as a less secure environment and therefore there is no false sense of security.

Is hacking into an aircraft's systems actually possible? :eek:

Please tell me no!

Please reread the 29 posts!:}

You're in luck. No.

I would imagine there are security concerns with regards to the ease that information on flights is available. But that is just information.

Actually hacking an aircraft systems, I don't really know what this guy is getting at. ACARS is a character based protocol, it just sends and receives text documents, not commands, just messages. Suggesting otherwise is like saying you can automatically make my phone dial a number by sending me an SMS. Not physically possible.

Fake ADSB targets is possibly a concern, as I think it is just data. I'm not sure, but I don't believe TCAS can be fooled as the range and bearing is internally computed by the interrogating aircraft, therefore only the altitude can be spoofed and the spoofer would need to be overflown by the aircraft to fool the aircraft into an RA.

It is indeed, totally possible! The SMS protocol was adapted so that people could send point to point text messages, but was originally designed into cell phones as a control protocol, a function it still also serves today. Most people don't know that things like your voice mail message indicator status or caller ID are carried as control messages via the same SMS protocol as text messages. They are just processed by the phone so you don't actually see them. Flaws in cell phone firmware have led to some exploits in the past, most notably, the iPhone: How To Hijack 'Every iPhone In The World' - Forbes.com These attacks were carried out by sending maliciously formatted SMS messages which the phone failed to reject (Apple was a bit of Newbie in the phone world at the time). It's the same situation here.

I wondered if that would come back an bite me :ouch:

After a bit of digging, this iPhone hack is not what is seems. It is basically a code injection exploit and requires the user to install an app. The security issue was that Apple allowed the app into the App Store without noticing the app's ability to receive and execute code that could possibly take control of your phone. The main stream news articles seem to be suggesting that it is possible to just send an SMS in the standard way, which is not the case.

Likewise, the reporting of this issue. This code injection isn't physically possible with ACARS and certainly is not the method talked about in that presentation anyway. That bloke is suggesting that you can generate false alerts, steer the aircraft and even turn it into a lawn dart with ACARS as it is, unmodified and using off the shelf gear. It isn't clear exactly how he demonstrated the vulnerabilities. Whatever, suggesting that he can remotely control the aircraft with ACARS or ADS-B is just looney tunes.

If you look at the pdf presentation he is saying that there are multiple ways to access ACARS on an aircraft if you know its address, ARINC and SITA make this a selling point. So getting ACARS messages up to the aircraft is simple.

He then uses standard hacking techniques like malformed messages, for esample an ACARS message that should have a character count instead provides a negative number or a ginormous number, he can do this because he is not trying to send ACARS messages he is trying to break the receiving software and he is not using an ACARS friendly transmission system. The computer that is running the ACARS software is _also_ the one in which a whole pile of other things run including the FMC, display processing, MCDU etc etc. So if he can make it run some exploit code by sending it a broken message that then allows him to upload some more code running at high authority, he has broken into the computer that is running around "80 - 100" of the major control applications of the aircraft.

Its all on the pdf slides.

Re Forbes' article, "How to Hijack 'Every iPhone iIn The World' ", here is Forbes' take on the present issue under discussion: Researcher Says He's Found Hackable Flaws In Airplanes' Navigation Systems (Update: The FAA Disagrees) - Forbes

The claim has been made that the author of this presentation could send radio signals to aircraft such that aircraft so hacked could be commanded to change direction, altitude and speed.

So far, no one has even outlined the steps and process let alone detailed the unbroken chain of technical events which would lead to an Android phone taking control of an aircraft beyond the crew's awareness or ability to counter the hack.

For the reasons I have posted above and in subsequent posts I suggest that the challenges are almost insurmountable when it comes to actually manipulating the flight controls as claimed by Mr. Teso.

That ACARS, ADS and CPDLC are potentially hackable is not in dispute. It is the extent of the threat that requires delineation, and I propose that such threats do not encompass actual control of the aircraft.

Hacking instrument displays such as Primary Flight Displays and Nav Data Displays with examples such as TCAS & GPWS indications still require assessment, resolution and action by crew members, and are not automatic responses by the autoflight systems. The implementation of providing uploaded data to make such a system "think" there is either an intruder (TCAS) or a legitimate EGPWS warning, is a herculian undertaking which would yield very little cockpit effect and zero aircraft effect. The "return" on the "investment" required just isn't there.

Route changes require manual executions using the FMS keyboard interface; no autoflight system changes altitude without manual input by the flight crew; speed changes are an autothrust function but inputs are reasonableness-checked by the FMS.

PJ2

There's a risk that the following quotations were reported in (InformationWeek | Business Technology News, Reviews and Blogs and Information for the World's Business Leaders - Forbes.com) incorrectly or may become superseded, the overall message though is quite clear; Honeywell, Rockwell-Collins, EAS and FAA aren't presently worried.

Honeywell spokesman Scott Sayres via phone
“If we talk very generically -- not just about Honeywell software -- PC FMS software is normally available as an online pilot training aid”
“In other words, what Teso did was hack a PC-based training version of FMS that's used to simulate the flight environment, not the actual certified flight software installed on an aircraft.”

Rockwell Collins
“Today’s certified avionics systems are designed and built with high levels of redundancy and security. The research by Hugo Teso involves testing with virtual aircraft in a lab environment, which is not analogous to certified aircraft and systems operating in regulated airspace.”

EASA spokesman Jeremie Teahan via email
“This presentation was based on a PC training simulator and did not reveal potential vulnerabilities on actual flying systems"
“There are major differences between PC-based training FMS software and embedded FMS software. In particular, the FMS simulation software does not have the same overwriting protection and redundancies that is included in the certified flight software”
“For more than 30 years now, the development of certifiable embedded software has been following strict guidance and best practices that include in particular robustness that is not present on ground-based simulation software”

FAA
“The FAA is aware that a German information technology consultant has alleged he has detected a security issue with the Honeywell NZ-2000 Flight Management System (FMS) using only a desktop computer. The FAA has determined that the hacking technique described during a recent computer security conference does not pose a flight safety concern because it does not work on certified flight hardware. The described technique cannot engage or control the aircraft’s autopilot system using the FMS or prevent a pilot from overriding the autopilot. Therefore, a hacker cannot obtain “full control of an aircraft” as the technology consultant has claimed.”

In my experience, simulations are the basis of extremely powerful techniques towards understanding the normal (expected) and emergent behaviour of any complex system, but it is usual to increase the throughput of test data by removing something. Once a cut-down simulation of a system has been produced, there is the need for extreme care in the usage of test results, as they can produce misleading positive and negative views of the real thing.

IMO PJ2, in particular, has correctly highlighted the irresponsibility of the presentation and subsequent re-broadcast of the work. In other fields, we know that, such presentation would not occur without peer review. It's actually a great shame that further output from the individual and their organisation may be devalued to some extent, as a consequence of this publicity.

I'm happy to fly alongside an Android and I'm happy to participate in a simulation, provided that it stays in the laboratory.

00, 01, 10 n 11

PJ2
If I have exploit code inside the Common Core System, then I put the characters I want into the FMS keyboard buffer storage area in the CCS followed by a return and it executes just as if the crew had entered it. Nothing would appear on the CDU. The slide he shows also indicates that the display outputs come via the CCS - so the displays could be made to show something entirely different to what the aircraft is actually doing. A hacker could reallocate all the crew accessible keys in the translate tables so that the crew would be unable to use the CDU or perhaps even on/off switches if they are actually 'software configured'.

Please forgive my ignorance but I thought aviation systems were among the most protected If not the most protected) on the planet!

Don't national aviation authorities insist on an aircraft's software systems impenetrable when they certify a new aircraft and then require it to continue being impenetrable?

Ian W, thanks for a detailed response. I can appreciate the potential vulnerabilities a bit more clearly. We'll let this "bake" for a while and see what next week brings. - PJ2