1. Attachments are working again! Check out this thread for more details and to report any other bugs.

Relay Attacks on Passive Keyless Entry and Start Systems in Modern Cars

Discussion in 'Gen 3 Prius Main Forum' started by observing, Jan 19, 2011.

  1. observing

    observing Junior Member

    Joined:
    Jul 3, 2009
    16
    0
    0
    Location:
    Baltimore, MD
    Vehicle:
    2010 Prius
    Model:
    II
  2. bisco

    bisco cookie crumbler

    Joined:
    May 11, 2005
    110,133
    50,049
    0
    Location:
    boston
    Vehicle:
    2012 Prius Plug-in
    Model:
    Plug-in Base
    this is a major problem and it turns out thousands of cars have been stolen using this method and shipped to south america where they are now being used as homes for people under 5' tall.
     
  3. 32kcolors

    32kcolors Senior Member

    Joined:
    Jul 14, 2009
    5,683
    953
    124
    Location:
    Redondo Beach, CA
    Vehicle:
    2010 Prius
    Model:
    V
    People will and can steal even the most well-protected cars so that's what insurance is for.
     
  4. jayman

    jayman Senior Member

    Joined:
    Oct 21, 2004
    13,439
    641
    0
    Location:
    Winnipeg Manitoba
    Vehicle:
    2004 Prius
    The problem here is that cars are expected to last a decade if not longer. Nobody will keep a computer or a wireless router a decade.

    The hack was demonstrated by Johns Hopkins and RSA Labs back in 2005

    About RFIDs and the TI DST - Car Thief Stoppers

    By 2008, the original RFID system was shown to be completely broken

    24C3 Mifare crypto1 RFID completely broken - Hack a Day

    As FPGA (Field Programmable Gate Array) devices become even cheaper and powerful, along with host devices like smartphones, we will see the pendulum swing from RFID helping secure our cars, to making them absolutely easy to steal

    Of course, there is no evidence of the theft. No broken windows, no busted steeering column, just the car mysteriously vanished and the insurance company denying a claim
     
  5. rogerSC

    rogerSC Member

    Joined:
    Jul 27, 2005
    170
    7
    0
    Yes, saw this article today. Most likely there's a business there in making attractive, practical RF shield devices for keyless entry fobs. If it could be made in a way that you could just reach in your pocket or purse and "unshield" it enough to work without having to take it out of your pocket, and then re-shield it when the car lock was open, that wouldn't be too bad (isn't that about the same thing as the non-keyless fobs, though?).

    Maybe they're already available, I haven't looked...this would be the Toyota recall, put your keyless entry fob in a metal case *smile*. My gas pedal is already shorter than it used to be.

    I'm having a hard time dismissing this with "that's what car insurance is for", myself.

    -Roger
     
  6. David Beale

    David Beale Senior Member

    Joined:
    Jul 24, 2006
    5,963
    1,985
    0
    Location:
    Edmonton Alberta
    Vehicle:
    2012 Prius
    If you can show them the two FOBs your insurance will cover the loss. You don't need a pile of broken glass.

    So a knowledgeable professional can steal your Prius. So what else is new. They can also use a flatbed car carrier. Or how about a tow truck with bogey wheels under the front wheels of your Prius?

    If you want a theft-proof car it's the one you haven't purchased yet.
     
  7. observing

    observing Junior Member

    Joined:
    Jul 3, 2009
    16
    0
    0
    Location:
    Baltimore, MD
    Vehicle:
    2010 Prius
    Model:
    II
    Some additional thoughts on the topic:

    * A quick scan of my favorite electronics part distributors (Mouser and Digikey) indicates that the bill of materials for a wireless configuration would be less than US$ 150 and could be built into two fairly small component packages. I don't know how that compares to the price of renting a tow truck, but I suspect it's competetive.

    * The parking garage scenario described in the paper seemed reasonable and likely. Based on admittedly rough system design considerations, the receiver component of the wireless system could easily be hidden out of sight or hidden on a person (if the thief isn't working alone). In any case, I would think it would draw less attention than the appearance of a flatbed carrier or tow truck and my assumption is that thieves don't like to draw attention to themselves.

    * A potential Mid-term Countermeasure that I didn't see mentioned in the paper could be the addition of a second access control for starting the car, such as a keypad on the dashboard. Starting the car would then require the fob and a pin/password. The relay attack would still get you in the car, but set off an alarm if the pin were not entered within a certain time period. I haven't checked whether these types of access controls are commercially available, but my assumption is that anything that causes a car thief to spend more time than he intended would be a significant deterrent.

    * There were a few comments about how insurance companies would react to a car stolen by this method. I'm guessing that insurance companies don't have a standard way of handling the report of a mysteriously vanished car, but I do remember that before "bump keys" became widely known, people were reporting that insurance companies were denying claims of illegal entry. Perhaps, insurance companies' responses to this attack will be similar. In the end, lock manufacturers were compelled to design bump-proof locks. If we're lucky, auto manufacturers were feel a similar compulsion.
     
  8. ETC(SS)

    ETC(SS) The OTHER One Percenter.....

    Joined:
    Oct 28, 2010
    7,855
    6,656
    0
    Location:
    Redneck Riviera (Gulf South)
    Vehicle:
    Other Non-Hybrid
    Model:
    N/A
    I don't think most auto insurers will make you produce keys to prove that your car was really REALLY stolen. In fact, I think that they would have to cover the loss even if you left your key-fobby-thingy in the car.
    Caveat: I've never had a car stolen...but I have a close relative that's been in the home/auto/life insurance biz for many years.

    Mostly, they'll just take the police report...wait a looooong time for the vehicle to be recovered, and if it's not, they'll (eventually) pay the loss. This is why car torching is such a common occurrence when slugs want to get out from underneath their monthly car payments.

    One thing is for sure. If your car ever does get stolen---it's private sale value will take a big ding once your VIN# gets attached to a police report, even if your kid just stole it for a quick spin around the block!

    As far as the "Relay Attacks on Passive Keyless Entry and Start Systems in Modern Cars'' worries....it's not much to worry about in Priuses, unless you're just looking for something to worry about. According to the Highway Loss Data Institute's report on the most stolen, broken into, or stripped American cars of 2010---we're nowhere near the top of the list...(We're probably like, #74---behind the Suzuki Kizashi...)

    Interestingly enough.....my GMC was number 6, and I sleep pretty well at night with my dumb old remote door lock/metal key start!
    TIFWIW..... ($0.02).... :D
     
  9. David Beale

    David Beale Senior Member

    Joined:
    Jul 24, 2006
    5,963
    1,985
    0
    Location:
    Edmonton Alberta
    Vehicle:
    2012 Prius
    Well, a friend had a Civic CRX Si stolen from his driveway. He had full replacement value insurance on it, but he did have to show them the two keys (to prove his son/friend/relative/etc) didn't take it. He got paid out in a week. No indication of how they did it. It was just -gone-.

    Do keep in mind the Prius is not an expensive car. Even the parts are not in great demand. Most thieves will be after your neighbors Lexus/Mercedes/Caddy/Lincoln/etc. Hondas are also in great demand for parts, because they are so popular for the hop-up crowd.
     
  10. 32kcolors

    32kcolors Senior Member

    Joined:
    Jul 14, 2009
    5,683
    953
    124
    Location:
    Redondo Beach, CA
    Vehicle:
    2010 Prius
    Model:
    V
    Correct.
     
  11. Judgeless

    Judgeless Senior Member

    Joined:
    Jun 30, 2009
    1,821
    256
    0
    Location:
    Cleveland, OH
    Vehicle:
    2010 Prius
    Model:
    IV
    Great find.
     
  12. pEEf

    pEEf Engineer - EV nut

    Joined:
    Sep 23, 2010
    720
    578
    3
    Location:
    Berkeley, CA
    Vehicle:
    2008 Prius
    Model:
    N/A
    There are a lot easier ways of stealing most cars (without towing) rather than the method described. The point of adding technology is to make it difficult, but it's never impossible.

    I don't think I've ever heard of a Prius being stolen, even though the technical capability exists, so I think we are pretty safe!

    I have a friend who has had all her windows broken by kids about 4 times now. That's probably worse!
     
  13. Judgeless

    Judgeless Senior Member

    Joined:
    Jun 30, 2009
    1,821
    256
    0
    Location:
    Cleveland, OH
    Vehicle:
    2010 Prius
    Model:
    IV
    I am sure you are right.
     
  14. Judgeless

    Judgeless Senior Member

    Joined:
    Jun 30, 2009
    1,821
    256
    0
    Location:
    Cleveland, OH
    Vehicle:
    2010 Prius
    Model:
    IV
    Not sure if this applies to a Prius.
     
  15. qbee42

    qbee42 My other car is a boat

    Joined:
    Mar 2, 2006
    18,058
    3,075
    7
    Location:
    Northern Michigan
    Vehicle:
    2006 Prius
    Here is what I don't understand about this form of attack: Most smart key systems, such as the one used in the Prius, use both RF (radio frequency) and LF (low frequency - essentially magnetic fields) for communication and detection. The RF is used for communication between the car and fob. LF is used for localization. The LF fields are generated by the coils in the car: one in each front door, one inside the car, and one at the rear for the three door system.

    RF has a relatively long range. If you press the the unlock button on your fob, it sends an command to the car via RF, which causes the door to unlock. This mode requires active operation by the owner, and is not susceptible to relay attack. A relay could extend the range, but you would still need to press the unlock button. Also, this would only unlock the car, not allow it to start.

    In smart mode, a beacon signal is used to tell the fob that it is close to the car. When the fob senses this beacon, it measures the LF field to determine the fob's orientation from the car and the distance from the car. It then encodes this information and sends it back to the car via RF. The car looks at this positional information and decides which door to enable for unlocking.

    In a similar fashion, the LF field is used to verify that the fob is inside the car in the drivers area before the engine is allowed to start.

    The RF signal is easy to relay. All you need is a receiver, a transmitter, and two antennas. This is what makes a relay attack possible. However, it isn't so simple with the LF field. Low frequency magnetic fields don't amplify and relay the same way as RF. For this to work, you would need to generate a phantom LF signal in close proximity to the fob, and then use the RF relay. You would have to do it twice with two different LF signals: once simulating the LF on the outside of the vehicle to allow the door to unlock, and a second time on the inside of the vehicle to start the car. This isn't a trivial problem to solve.

    So how did the relay attack work on some cars? It's likely these cars use a simplified smart key system that doesn't include the LF oscillators. It's just like mechanical locks: some are cheap and easy to break; others are very secure. Some are much better than others.

    Tom
     
  16. macman408

    macman408 Electron Guidance Counselor

    Joined:
    Mar 21, 2010
    1,179
    367
    1
    Location:
    California
    Vehicle:
    2010 Prius
    Model:
    V
    I'd recommend not being so confident it can't be broken - I have no doubts it can be. It's not terribly likely that lots of people will start doing it, of course. It's more likely that somebody would try an attack that doesn't require a relay, which also might be possible. A relay attack is just a pain.

    Car manufacturers haven't had to think much about security - like what happens if somebody figures out a way to hack the Nav unit via Bluetooth? Or the TPMS wireless links? Some of these things may not have very well-thought-out security. Sometimes, that's just because, well, 'who cares if somebody figures out how to make your TPMS light turn on?' But cars haven't been something that people have traditionally tried to write viruses and other malware for. And maybe they never will. But it doesn't mean it can't be done. It's just a matter of return on investment - if you have to spend $100,000 to steal a $20,000 car, or invest months of time to do it, it's probably not worth it to most thieves.
     
  17. RobH

    RobH Senior Member

    Joined:
    Sep 18, 2006
    2,369
    980
    70
    Location:
    Sunnyvale, California
    Vehicle:
    2006 Prius
    My guess is that the low frequency signal is just another radio signal. The low frequency (134.2 KHz) and small antennas make for the short range. The car-to-keyfob information is something like "this is the driver's door oscillator looking for a key for car X". When a keyfob hears the signal, it compares its car number with the transmitted value. If there is a match, then the keyfob replies "I hear the driver's door oscillator and here is my current rolling code and encrypted identification". Each oscillator could be transmitting a different message that identifies its location on the car.

    I know that the keyfob signal uses rolling code and encryption. I don't know what the car-to-keyfob data is. Since the car-to-keyfob data can be interpreted by all keyfobs, it probably isn't encrypted. In fact it may not even be varying. I wonder if you recorded the car-to-keyfob signal, would replaying that signal cause the keyfob to reply? My guess is yes.
     
  18. observing

    observing Junior Member

    Joined:
    Jul 3, 2009
    16
    0
    0
    Location:
    Baltimore, MD
    Vehicle:
    2010 Prius
    Model:
    II
    According to the paper, the authors up-mixed the car's 130KHz beacon signal to 2.5GHz, which is easily transmitted to a receiver near the fob and down-mixed back to 130KHz. The fob then transmits its long-range RF signal directly to the car. The assumption is that the person carrying the fob will remain in proximity to the receiver long enough for the thief to open the car door, get inside, and press the power on button.
     
  19. qbee42

    qbee42 My other car is a boat

    Joined:
    Mar 2, 2006
    18,058
    3,075
    7
    Location:
    Northern Michigan
    Vehicle:
    2006 Prius
    Yes, but this wouldn't work with the LF position sensor. Any idea how you would defeat that?

    Tom
     
  20. RobH

    RobH Senior Member

    Joined:
    Sep 18, 2006
    2,369
    980
    70
    Location:
    Sunnyvale, California
    Vehicle:
    2006 Prius
    Unlocking the door and hitting the Power button are two distinct processes from a SKS point of view. To unlock the door with SKS from the outside, the car has to be powered off and doors locked. The keyfob has to be detected within about 3 feet of the outside of the door. When this condition has been met, then touching the sensor on the door handle triggers a door unlock. It also turns off the the outside oscillators, and turns on the inside oscillators. In order to enable the Power button, the keyfob then has to be detected by the inside front oscillator. This is a second detection, at a different location.

    I've observed this behavior with a low-band receiver that can tune to the low frequency oscillators. Actually, I have to tune it to the 2nd harmonic at 265 KHz, but the data is there. The outside oscillators are on while the car is turned off and locked. As soon as the car is unlocked, the outside oscillators turn off and the inside ones are on. When the car is turned on, all of the oscillators are disabled. There are a few more conditions that cause the oscillators to turn on, such as opening the door with the power on.