4 RFID Security Settings and Assumptions
To address the security risks of low-cost RFID tags, we will first state a set of assumptions
about the operation of the system. Assuming a minimalist approach, tags will be
passive and provide only simple read-only identification functionality.We will arbitrarily
assume our tags contain a few hundred bits of storage and have an operating range
of a few meters.
In 2003, cost requirements dictate that low-cost tags may have 200-2000 gates available
for security. This is far below what is feasible for standard public-key or symmetric
encryption, including efficient algorithms such as NTRU or TEA [11, 32]. Furthermore,
performance requirements dictate that at least 100-200 tags must be able to be read
each second, which limits the clock cycles available for security protocols. Power consumption
may also be a limiting factor, although highly dependent on the particular
implementation.
We assume tag memory is insecure and susceptible to physical attacks [29, 31] revealing
their entire contents. This includes a myriad of attacks such as shaped charges,
laser etching, ion-probes, TEMPEST attacks, clock glitching and many others. Fortunately,
these attacks require physical tag access and are not easily carried out in public