Researchers successfully hack into automobiles using passive keyless systems

Jan 07, 2011 by John Messina weblog
Attacker starts a car using an antenna. A signal from the car is transmitted to a computerized key, which is tricked into enabling the engine ignition.
Credit: ETH Zurich

(PhysOrg.com) -- Researchers at a system security group at ETH Zurich in Switzerland were able to access ten automobiles from eight manufactures and drive them away.

A passive keyless entry system is activated when a wireless key is within a few meters of the correct automobile and detects a low power signal from the vehicle. The wireless key then sends a command that opens the vehicle and starts the ignition.

Srdjan Capkun, an assistant professor of computer science, along with his colleagues, managed to intercept and relay signals from the vehicle to their wireless keys. They could have also relayed the signal from their wireless key back to the vehicle, but choose not to because the key can transmit its signal up to 100 meters (approx 328 feet). Their attack proved successfully no matter what cryptography and protocols the key and vehicle used for communication.

The attack was carried out by using a pair of antennas to transmit signals from the vehicle to the wireless key when the key was far away, tricking the vehicle into opening and starting. One needs to be very close to the automobile while the other needs to be within eight meters of the key.

Most relay attacks require the signal to be converted from analog to digital and back to analog again causing delays in microseconds. These delays can cause the vehicle not to open or start. This was circumvented by keeping the signal in analog format, cutting the delay down to nanoseconds.

There’s not too much owners can do to protect themselves except for maybe shielding their wireless key when leaving their vehicle. Capkun says manufactures will need to add secure technology that allows the vehicle to confirm that the wireless key is close by. The researchers are actively working on protocols that would make this happen.

Explore further: Singing the same tune: Scientists develop novel ways of separating birdsong sources

More information: Via: Technology Review

Related Stories

Beware of Hackers Controlling Your Automobile

May 18, 2010

(PhysOrg.com) -- A team of researchers led by Professor Stefan Savage from the University of California, San Diego and Tadayoshi Kohno from the University of Washington set out to see what it would take to ...

Jaguar Leaps into Luxury of Bluetooth Wireless Technology

Mar 14, 2005

Motorola, Inc. and Jaguar announced the availability of the Jaguar Bluetooth system – a new hands-free in-vehicle communication system using Bluetooth wireless technology – across all Jaguar 2005 model year vehicles. Once a J ...

The vehicle industry must have a rethink

Sep 21, 2009

Despite the fact that the vehicle industry is currently described as an industry in crisis, there are high hopes that the development of future IT services for vehicles will drive development. However, for new technology ...

Image: To the rescue

Nov 17, 2010

This unique, close-up view of the X-38 under the wing of NASA's B-52 mothership prior to launch of the lifting-body research vehicle was taken from the observation window of the B-52 bomber as it banked in ...

Recommended for you

Taking great ideas from the lab to the fab

1 hour ago

A "valley of death" is well-known to entrepreneurs—the lull between government funding for research and industry support for prototypes and products. To confront this problem, in 2013 the National Science ...

User comments : 30

Adjust slider to filter visible comments by rank

Display comments: newest first

Quantum_Conundrum
1.5 / 5 (8) Jan 07, 2011
Their attack proved successfully no matter what cryptography and protocols the key and vehicle used for communication.


Well duh. The fake signal is an exact copy of the original signal, so it's automatically got the code build into it. You don't need to know how to code or decode the data, you just need to be able to clone the signal. The length of the password doesn't matter either.

I still think this should make any and all wireless communications "hackable," because anyone at all can intercept the signal, they just need to figure out how to decode the signal. If they are using the same equipment and software, the algorithm is already build in...
dtxx
4.6 / 5 (9) Jan 07, 2011
NO. What you are describing is a replay attack, and there are plenty of ways it is mitigated. "Just need to figure out how to decode the signal." Are you serious with that? I think you are, and that actually makes it even funnier than if you were joking.
Pyle
3.3 / 5 (3) Jan 07, 2011
Something as simple as a chime on the key to alert the owners the key was activated might help, but that isn't really an effective solution.
Truthforall
3 / 5 (1) Jan 07, 2011
Putting a switch on the wireless key would help.
Press once to activate and will turn off after a short time unless docked to the car.
This hack will work for burglars. Car thief would find himself trapped when he drive out of range.
CSharpner
4.5 / 5 (2) Jan 07, 2011
The flaw in this system is that it happens without owner intervention. Just like secure OS's prompt the user before a malicious program is allowed to perform a secure action, the key should prompt the owner for permission to respond to the unlock or start request. Even better, let the owner initiate the request by pushing a button on the key.

Additionally, the car can time the response and if it responds too late then it should reject the request. If it can't respond in the amount of time it takes the radio signal to travel out a couple yards and back + minimal processing time in the key, then it should assume the key is out of range.

Just because the car receives the proper response doesn't mean it's a "good" response. It must also respond quick enough.

Both the owner prompt and the timing should be implemented.
gurloc
4 / 5 (2) Jan 07, 2011
The passive key concept itself is idiotic. To protect against a relay attack you would need to use tight timing cuts to determine the keys position. That would require fast electronics in the key itself pushing up its power requirements and either shortening its battery life or forcing the use of a larger battery.
CSharpner
4.5 / 5 (2) Jan 07, 2011
I still think this should make any and all wireless communications "hackable," because anyone at all can intercept the signal, they just need to figure out how to decode the signal. If they are using the same equipment and software, the algorithm is already build in...

Not quite. "Figuring out how to decode the signal" is a much larger gargantuan task than you can imagine. Even if you know the encryption algorithm, you still have to provide the right key. A brute force attack on a simple 128 bit key would be impossible because just the amount of energy for all the bit flipping inside the microprocessors alone to run through, on average half the keys to break it requires more energy than burning through ALL energy in the universe PLUS converting ALL mass in the universe to pure energy and consuming that... even then, you wouldn't be anywhere close to cracking it. Then, the amount of time involved is greater than the age of the universe.
CSharpner
5 / 5 (3) Jan 07, 2011
As for cracking any wireless using this technique... Absolutely NOT. Standard wireless doesn't work the same way. The vehicle key hack works by essentially extending the range of the network. They're not hacking the network. They're waiting for legitimate network traffic to start the car, then they steal the car. At no point do they ever decode the wireless signal. To "hack" a wireless network, the point is to decrypt the signal so you can eavesdrop on the communication to either steal data and/or take control of the system. This is radically different than the car theft scenario.

CSharpner
"Hacking" since 1982
Hacking: To learn as much about a system as possible. To use that knowledge for good or evil does not determine whether it's "hacking". Evil hackers get the news. Good hackers get the jobs.
CSharpner
5 / 5 (2) Jan 07, 2011
To protect against a relay attack you would need to use tight timing cuts to determine the keys position. That would require fast electronics in the key itself pushing up its power requirements and either shortening its battery life or forcing the use of a larger battery.

That's why you implement the timing in the vehicle and not the key, ASSUMING that "tight" timing requires more power. It doesn't necessarily take more power, just very small electronics, which usually require less power. Anyway, you'd want to offset as much processing as possible to the car, obviously, for power reasons. Also, if the key has a button and is never "on" until a button is pushed, it's battery should last a long time.
Quantum_Conundrum
1.7 / 5 (6) Jan 07, 2011
somehow, the fact that major websites are hacked all the time makes me suspect it isn't that hard after all.

there are 3.4E38 combinations, but we can typically eliminate most of them because people tend to use idiotic passwords that are nowhere near true randomness.

the problem with true random passwords is nobody can remember them, so you end up having to write them down somewhere, which then makes you more vulnerable than if you had an easy to remember password anyway.
CSharpner
5 / 5 (4) Jan 07, 2011
somehow, the fact that major websites are hacked all the time makes me suspect it isn't that hard after all.

Well, yes and no. It depends on what you're trying to hack:

1. The encryption tech.
2. Social engineering a user's password.

In a product like a key for a vehicle, there's usually not a user created password, so it can be random and can be very large, making it impossible to break.

On the other hand, web sites are hacked not because anyone ever breaks the encryption, but because they work around it by finding parts that aren't protected or guess user passwords like "sex" (the most common pw) or "password" (2nd most common). There are also holes in operating systems and web browsers and email programs and web server software, all of which are exploited to sneak in without ever breaking the encryption. And let's not forget bad programming practices that allow sql injection, xss, etc. So, yes, some sites are easily hacked, but never by breaking the actual encryption
CSharpner
5 / 5 (3) Jan 07, 2011
Not to mention that hacking a web site doesn't have anything to do with how hard or easy it is to hack a wireless connection. Assuming a long and "random" key, it's very difficult. WEP is easily hacked though... within seconds. It was a flawed system. WPA2 is a better one, but you, of course, need a strong key and you have to memorize it. It CAN be done. I do it. I NEVER write down any of my passwords... EVER. If I ever had to reset my password on a website and they e-mail me a new one, I immediately change it. Common sense and best practices go a long way. But, you're right in a sense... any system with poor passwords can be broken easily with a dictionary attack. If you don't secure your entire attack surface, you're vulnerable.

Anything can be broken, but good practices can make it too difficult for a malicious user to try.
antialias
3 / 5 (1) Jan 07, 2011
The fake signal is an exact copy of the original signal, so it's automatically got the code build into it.


That particular strategy doesn't work because basically all car manufacturers use rolling keys.

the problem with true random passwords is nobody can remember them

There are mnemonics you can use to generate relatively good keys. E.g. take a number you know well (lets say: 773265). Then take the letters of the site you want to generate a password for (lets say: physorg.com)
Pick the letters in the positions of the digits and disperse the number in any place you can remember (e.g. one letter, 3 numbers, 2 letters, 2 numbers, 3 letters, one number). Result:
.773.y26hgr5

- Strong password
- Unique for each site you access
- Easy to remember
- if you do forget it it's easy to reconstruct
Quantum_Conundrum
4.3 / 5 (3) Jan 07, 2011
A brute force attack on a simple 128 bit key would be impossible because just the amount of energy for all the bit flipping inside the microprocessors alone...requires more energy than burning through ALL energy in the universe...


this is definitely wrong man. You can approximate the power flux of the sun by just applying the solar constant at 1 A.U., i.e. 1350 watts/m^2, to a uniform shell at 1 a.u. radius.

The power of the sun is 3.82E26 watts.

Now my PC has a 500watt power supply, most of which isn't even being used. I have 4 cores at 2.8ghz, which gives total cycles of 11.2 billion per second. Now even if the whole 500 watts was for processor and memory alone, that would be 22.4 million clock cycles per watt.

Which means that one second worth of the sun's power is enough energy to do 8.512E33 cycles on my PC. Now it would never last long enough to do that...

2^128 = 3.4E38.

11.1 hours worth of the sun's total power is enough to do that...
Pyle
4.5 / 5 (2) Jan 07, 2011
@QC: good call on the hyperbole. Shame on you CS.

@antialias:
In your first part you miss the point. The signal is constantly broadcast from the key. The hacker just relays the message with antennae so the car thinks the key is close. Rolling doesn't matter because the key is really talking to the car, just through an intermediary.

Second part of your post is decent advice.
PinkElephant
5 / 5 (1) Jan 07, 2011
My work building uses a security card access scheme, where the card must be within just a couple of inches from the reader before it can be read (I'm guessing it's based on some resonator scheme, where the time delay between sender and receiver must be extremely short.) Anyway, point is my security card doesn't even have a battery; it's a passive device and it's energized (probably similar to RFID) by the reader.

Similarly, if they tighten the radius for the smart key activation from 8 meters down to, say, 2 feet, plus require the car to "ping" the key every few seconds in order to continue driving, then it would render this attack moot. Of course, it would entail building in several antennae into the car, to essentially surround the cabin. But you couldn't very easily steal my car if your scheme required you to keep a relay antenna within less than 2 feet of my body, without me either noticing it or accidentally walking out of range.
DamienS
5 / 5 (1) Jan 07, 2011
The passive key concept itself is idiotic.

I'd like to second that!
Researchers at a system security group were able to access ten automobiles from eight manufactures and drive them away.

Is there an implication there that some vehicles couldn't be hacked? If so, how many and why were they more resilient?
antialias
not rated yet Jan 07, 2011
In the link to the article it says that they just tested on 10 vehicles from 8 manufacturers and were able to drive all of them away.

@Pyle: You're right. I was mistakenly of the opinion that they were talking about normal (button activated) electronic car keys. The linked article makes the technology/method much clearer.
trekgeek1
not rated yet Jan 07, 2011


this is definitely wrong man. You can approximate the power flux of the sun by just applying the solar constant at 1 A.U., i.e. 1350 watts/m^2, to a uniform shell at 1 a.u. radius.


Good catch. That raised my eyebrow too. Even simpler is to realize that it takes much less than a Joule to flip a bit and there are more than 2^128 Joules of energy in the universe.
CSharpner
5 / 5 (1) Jan 07, 2011
Qc,

Which means that one second worth of the sun's power is enough energy to do 8.512E33 cycles on my PC. Now it would never last long enough to do that...

2^128 = 3.4E38.


Excellent QC! I gave you a 4. I should have done the math before quoting what I saw on a show 10 years ago. He may have said a 1024 bit key (I'm too tired to do the math at the moment... could have been bigger). Anyway, the point is the same... not really plausible to crack.

BTW, one CPU cycle isn't nearly enough to perform a check on a key... most likely 1000 or so... maybe more depending on the algorithm and the quality of the programmer.

Pyle,
@QC: good call on the hyperbole. Shame on you CS.


Ooh! Harsh. True, but harsh. I second Pyle's note. Good call Qc! Now here I go walking the hall of shame. LOL! Everyone should ignore all the other cool stuff I wrote and focus on the insignificant point ;)

But seriously, good call QC.
DGBEACH
not rated yet Jan 07, 2011
rolling-code keyfobs will "encrypt" the ID transmitted according to a preset "key". Listening to their transmissions will quickly allow the hacker to figure it out...the equivalent of WEP encryption...wireless security is only a myth, it doesn't exist!
dtxx
2 / 5 (1) Jan 07, 2011
C#ner,

Hacking doesn't necessarily have anything to do with information systems, but I get the feeling you have at least some knowledge in that area. If so, you should know that the way websites are hacked these days usually has absolutely nothing to do with brute-forcing any passwords. Is it almost always the result of a) an exploit present in the OS or an application b)malware, such as a trojan or keylogger, c) insecure defaults or options selected by the sysadmin to make his life easier, or d) (as you mentioned) Social Engineering.

You are giving a bit too much credit to what QC says, and I'm surprised you let it pass. He has repeatedly shown he has absolutely no clue when it comes to computer science.
dtxx
not rated yet Jan 07, 2011
rolling-code keyfobs will "encrypt" the ID transmitted according to a preset "key". Listening to their transmissions will quickly allow the hacker to figure it out...the equivalent of WEP encryption...wireless security is only a myth, it doesn't exist!


You DO know that WEP was INTENTIONALLY created in a flawed state, right? Of course you don't. Feel free to respond, but I don't think I can take anything you say on this subject seriously after your last comment.
VOR
not rated yet Jan 07, 2011
Call me old fashioned, but I prefer a real ignition key with wireless entry fob. (The key also has a passive matching component for the ignition so you cant simply copy the key).
Kedas
not rated yet Jan 08, 2011
Didn't Knight Rider have fingerprint check.

Anyway add a hardware piece in it, they will always be able to break it if you keep it wireless.
an other idea: A weight check of the person sitting in the driver seat, if wrong she/he has to enter a (voice) password.

It would also stop car-jacking.
Au-Pu
not rated yet Jan 08, 2011
Ten years ago I took two inventors to General Motors in Melbourne Australia with a foolproof security system.
One that was electronic and could not be hacked by anyone.
They were told by GM that they were ten years too early.To go away and come back when the industry was ready for their product.
GM didn't want to lose all the theft replacement market to their competitors.
What do these guys do with their system?
They sell it as a garage door security system.
Whether people have code grabbers or not simply does not matter.
All an owner needs to do is to press "lock" and wait 1 second. If a code grabber wants to try to open the car he must do so within that second otherwise it cannot be opened other than by the key that locked it.
I will be contacting them and telling them there could be interest in their product, at last.
Skepticus
not rated yet Jan 08, 2011
This just shows that you can't have everything. Auto door openners, auto ignition, auto seatbelts,..auto car theft. I'd go for a locating becon for the car with palm/iris scanner at the door.
CSharpner
not rated yet Jan 08, 2011
Dtxx,
you should know that the way websites are hacked these days usually has absolutely nothing to do with brute-forcing any passwords.

Of course. That's why I said the following that makes that abundantly clear.
On the other hand, web sites are hacked not because anyone ever breaks the encryption, but because they work around it by finding parts that aren't protected or guess user passwords like "sex" (the most common pw) or "password" (2nd most common). There are also holes in operating systems and web browsers and email programs and web server software, all of which are exploited to sneak in without ever breaking the encryption. And let's not forget bad programming practices that allow sql injection, xss, etc. So, yes, some sites are easily hacked, but never by breaking the actual encryption

CSharpner
not rated yet Jan 08, 2011
Dtxx,
You are giving a bit too much credit to what QC says, and I'm surprised you let it pass. He has repeatedly shown he has absolutely no clue when it comes to computer science.

It's true. He has little, if any, knowledge about cs, but I give the man credit when he gets something at leat partly right. My statement about using up all the energy and mass in the universe to decrypt a 128 bit key was wrong on a cosmic scale and he caught it. Good call for him and I give him credit for it as he deserves in spite of the fact he grossly underestimated how many CPU cycles he thought it took to process one key. His main point was right on: There's plenty of energy to do it. He was right and I was wrong... simple as that. I'm not the kind of person to downvote everything from someone, even when they're right. QC does say a lot of stuff that's either wrong or that I disagree with, but he says stuff that's right sometimes too and he deserves credit when he does.
dains
1 / 5 (1) Jan 08, 2011
*whup nm, ninja'd by Pyle.