1
$\begingroup$

We're working with a microcontroller that is not secure enough to store secrets. The ROM could be dumped in a properly equiped lab.

This is why we considered integrating a certified Secure Element that would store the secrets we need for our infrastructure (some keys and certificates). This way the secret would be stored in a valid device regarding security standards.

However, in this new architecture, the Secure Element would be deported from the host microcontroller, which requires to secure the communication between the SE and the Host.

Yet, all the current authentication methods we know require to store something (a secret, a key a certificate), what we cannot do because the microcontroller is not secure enough...


That leads to two questions:

  • What is the purpose of integrating a Secure Element or Secure Access Module to address the needs of secret storage for an infrastructure if it requires another secret for connection?
  • Is there any authentication method that would not require secret, but would garantee anyway identities?

Many thanks

$\endgroup$
  • $\begingroup$ your concern about dumping ROM is basically ubiquitous. I can decap ICs. The only thing that you can actually use to store data securely is floating-gates that use current mode, and as far as I know no one outside of a few academic groups use them. If you have a MCU, you could look charge the IO pin capacitance and then read in a loop and count how long it takes for a "level change" due to the mismatch on the driver and read-in circuit. $\endgroup$ – b degnan Feb 18 at 12:04
2
$\begingroup$

What is the purpose of integrating a Secure Element or Secure Access Module to address the needs of secret storage for an infrastructure if it requires another secret for connection?

One can be quite sure that the genuine Secure Element is involved in a communication that it certifies. If the Secure Element is BGA-mounted and/or epoxy-bounded to a PCB, one can be reasonably confident that this PCB is involved, or that adversaries spent significant effort to workaround that.

Is there any authentication method that would not require secret, but would guarantee anyway identities?

Not in traditional cryptography, including asymmetric and post-quantum.

Physically Unclonable Functions aim at something close: the PUF is supposed to be impossible to analyse to a degree allowing its successful cloning. One should be reasonably sure that a PUF was involved in a communication that it certifies. If the PUF is integrated on a silicon die, that die was involved too. That's the general idea when the PUF aims at preventing unauthorized production of the die from unmodified photomasks (stolen or reverse-engineered), and/or make the die unclonable.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.