Slashdot Mirror
How To Hijack a Drone For $400 In Less Than an Hour
Trailrunner7 writes "The skies may soon be full of drones – some run by law enforcement agencies, others run by intelligence agencies and still others delivering novels and cases of diapers from Amazon. But a new project by a well-known hacker Samy Kamkar may give control of those drones to anyone with $400 and an hour of free time. Small drones, like the ones that Amazon is planning to use to deliver small packages in short timeframes in a few years, are quite inexpensive and easy to use. They can be controlled from an iPhone, tablet or Android device and can be modified fairly easily, as well. Kamkar, a veteran security researcher and hacker, has taken advantage of these properties and put together his own drone platform, called Skyjack. The drone has the ability to forcibly disconnect another drone from its controller and then force the target to accept commands from the Skyjack drone. All of this is done wirelessly and doesn't require the use of any exploit or security vulnerability."
In TFA he is hacking a Parrot AR wifi drone. If Amazon ever gets off the ground (ahem) with their drones, they will likely be autonomous, using GPS to guide them to their location. Monitoring and flight plan changes would likely occur by satellite as well. That's not to say that they are immune from attack, but none of the types of drones described in the summary (law enforcement, intelligence agencies, Amazon) are going to be susceptible to his attack.
"All of this is done wirelessly and doesn't require the use of any exploit or security vulnerability"
"...detects the wireless signal sent out by a target drone, injects WiFi packets into the target’s connection, de-authenticates it from its real controller and then authenticates it to the Skyjack drone"
Uhh... for what definition of "security vulnerability" is this not a "security vulnerability"?
All of this is done wirelessly and doesn't require the use of any exploit or security vulnerability.
Between me and the author of this sentence, I think we have two different definitions of "security vulnerability".
For something like Amazon's purported drones... all you'd have to do is to hardcode the delivery address and HQ into the drone before flying, and make sure it doesn't accept any incoming signals by turning the wireless off. Now, if we want to talk about trying to get the drone's GPS systems confused, that would be something else! (Actually I'm still wondering if the drone would be smart enough to land on pavement or miss entirely and drop packages on a customer roof or balcony.)
READY.
PRINT ""+-0
While pro-grade multicopters like those to be deployed by Amazon operate at 2.4GHz, they do not use WiFi as their radio system! Typically, these multicopters are fitted radio systems such as Futaba, JR, Spektrum or 9X, and therefore Skyjack will not be able to take them down.
w00t
a truck driver
Ok, so hang on, In a previous life as a military contractor, I used to do this with 1980's technology. This (TFA) sounds like a cheap, brute force approach, that actually works fairly well. You overwhelm the subject with a much stronger signal, and depend on the receiver's automatic gain control to limit the amplitude, putting the "real" control signal down in the noise. You then have the drone's full attention.
The usual countermeasure is to encrypt the control signal. Then, you can still do a DOS (in today's terminology), but you can't get the drone to obey your commands.
The counter-counter measure to this is to break the encryption so you can control the craft. Flash back to those supercomputers that hobbyists were building by clustering lots and lots of game consoles. Just saying'.
Then, there's counter-counter-counter measures like hopping between frequencies and so forth, but for every technique there's a counter-technique, and I suspect computers have gotten fast enough to analyze tricky incoming signals and mimic them fairly quickly.
Someone brought up GPS -- Amazon's little copters can't be hacked because they're autonomous, using GPS for navigation. Well guess what -- GPS is just another signal. As we learned in the middle east, it is possible to spoof those signals and get a drone to land in a place it didn't expect.
The counter to *that* is inertial guidance. But realistically, Amazon and most government agencies probably won't have the budget for that.
Optical guidance? (and optical surveillance in general) Green lasers with automated tracking and aiming triangulating by noise, or emitted RF, or visual recognition. Anyone with robotics experience should be able to at least theorize a solution.
Wow, the next few years are going to be *fun*.
Oliver's law of assumed responsibility: If you're seen fixing it, you will be blamed for breaking it.
If you like your drones... You may keep your drones...