in the case of a wireless realization), the additional delay added by the relay attack is proportional to the speed of the wave propagation in the cable and the length of the cable. For a standard RG 58 coaxial cable, the wave propagation speed in that cable is equal to 2/3 of the speed of light in vacuum (that we denote by c). Therefore assuming that the UHF reply propagates at the speed of light in vacuum, the relay with a 30 m long cable adds 30/c + 30/(2c/3) = 250 ns of delay to the measured round-trip time between the car and the key.
Thus, if the round-trip time measurement in the dis- tance bounding implementation shows a variance higher than 250 ns then it will be impossible to detect the above described attack. If this variance is few orders of magnitude smaller than the delay introduced by the relay then the ver- ifier will be able to deduce the response time of the key and therefore be able to compute the distance to the key reliably. Given that the maximum distance at which the key should be able to open the door (without action from the user) is at most 1 m, the maximum standard deviation of the measured round-trip time should be less than 2/c = 6 ns.
One recent implementation of RF distance bounding  showed that the processing time of the prover (key) can be stable with a rather small variance of 62 ps. This suggest that current and upcoming distance bounding implementa- tions will be able to meet the PKES requirements.
Sketch of the Solution
PKES system based on RF dis-
tance bounding would work in the following way. When the user approaches the car, the key and the car perform a secure distance bounding protocol. If the key is verified to be within 2 m distance, the car would unlock and allow the user to enter. In order to start the car, the car will verify if the key is in the car. This can be done using a verifiable multilateration protocol proposed in , which allows the car to securely compute the location of a trusted key. Ver- ifiable multilateration requires that at least three verifying nodes are placed within the car, forming a verification tri- angle, within which the location of the key can be securely
7 Related Work
Low-Tech ttacks on Car Entry and Go Systems Low- tech attacks such as lock-picking physical locks of car doors or using hooks can be used to open a car. The hook is pushed between the window and the door and the thief tries to open the door by hooking the lock button or command. However, these low-tech attacks are less reliable on new car systems or when an alarm system is present. Lock-picking also leaves traces which can be analyzed by a forensics in- vestigator .
significant amount of research
remote key entry systems such as Keeloq [25, 33, 13], DST . Vulnerabilities are often the consequence of
publicly reviewed by the community or side channel weak- nesses. Consequently, manufacturers are moving towards more secure and well established ciphers (e.g., tmel docu- mentation recommends ES ). However, solving such issues by moving to the best cipher to date will not solve physical-layer relay attacks. The relay attack is indepen- dent of the cipher used; no interpretation or manipulation of
the data is needed to perform a relay attack.
Jamming and Replay
well known attack against key-
less car opening systems is to use a simple radio jammer. When the user step away from his car he will push the key fob button to lock the car. If the signal is jammed, the car won’t receive the lock signal and will therefore be left open. If the car owner did not notice that his car didn’t lock, the thief will be able to access it. However a jammer can not help a thief to start the car. nother related attack is to eavesdrop the message from the key fob and replay it (e.g., using on a fake reader/key pair). Standard cryptographic protocols using a counter or a challenge-response technique
provide defense against message replay.
Major electronic parts suppliers provide
components for passive keyless entry systems [29, 44, 32, 30], those components are then used by various car man- ufacturers. lthough variations exists in the protocols and cryptographic blocks (Keeloq in , TI DST in , ES in ), all manufacturers provide systems based on the same combined LF/UHF radio technology as we discussed
in Section 2.
Therefore, those systems are likely to be im-
pacted by the attack we have presented.
ttacks on Keyless Systems The closest work to our in- vestigation can be found in [6, 7]. The authors perform se- curity analysis of Keyless Car Entry systems including relay attacks. While the performed analysis identifies the relay problem, the proposed relay attack consists of two sepa- rate UHF relay links to relay messages in both directions. The proposed abstract setup has the problem of creating a feedback loop as the car will also receive the relayed sig- nal from the second link. We show that such a realization is not needed in modern PKES systems and demonstrate it experimentally. Moreover, the authors do not provide nei- ther hardware design, nor practical implementation of the attack. Finally, no adequate countermeasures are proposed.
Some practical attacks on PKES systems have been re- cently reported . However, no detailed information is available and it is not possible to understand the details of