A User Centric Security Model for Tamper-Resistant Devices

8.2 Smart Card Runtime Environment
As noted, the reason for the success was the buggy JCVM. Their results were based on
eight dierent smart cards and they only managed to attack one of them, as the other
smart cards had eective runtime protection mechanisms. Similar results were also shown
by Lanet and Iguchi-Cartigny [195]. Sere et al. [216] use the similar attack of modifying the
bytecodes to gain unauthorised access or skip security mechanism on a platform. However,
Sere et al. relied on fault attacks to modify the bytecodes rather than modifying them
o-card as done by [190, 195, 215]. This way, Sere et al. managed to bypass the on-card
bytecode verication. A countermeasure to this attack provided by Sere et al. relied on
tagging the bytecode instructions with integrity values (i.e. integrity bits) and during the
execution, the JCVM checks these bits and if it fails, the execution terminates.
In 2010, Barbu et al. [194] along with Vétillard and Ferrari [198] used a similar attack
methodology to Sere et al. [216] that later came to be known as combined attacks. Later,
the combined attack technique was extended to target various components of JCVM in
[217][220]. These attacks are signicant; nevertheless, they require the loading of an
application designed specically to accomplish the attack goals. Therefore, such attacks
are not practical to some extent in the ICOM; however, due to the open nature of the
UCOM such attacks become a real concern.
In this section we glanced over the attack techniques proposed in the literature that specifically
target the SCRT. The discussion is by no means exhaustive but it introduces the
challenges faced by the UCTD runtime environment. Before we move to discuss the protection
mechanism, we rst discuss the fault attacks in some detail in next section.
8.2.2.1 Fault Attacks
The aim of an adversary during a fault attack is to disrupt the correct execution of an
application by introducing errors. These errors are usually introduced by physical perturbation
of the hardware platform on which the application is executing. By introducing
errors at a precise instruction, an adversary can circumvent the security measures implemented
by the runtime environment. Possible types of faults an adversary can produce are
described as below:
1. Precise bit error: In this scenario, an adversary has total control over the timing and
locations of bits that he wants to change.
2. Precise byte error: This scenario is similar to the previous one; however, an adversary
only has the ability to change the value of a byte rather than a bit.
3. Unknown byte error: An adversary has no control on the timing and byte that it
modies during the execution of an instruction.
193