One Laptop Per Terrorist

[Markus Ottela]

Now that I think of it, SW implementations with the CEV version that
cascades symmetric ciphers are very very slow if SoCs such as RPi are
used. OTP and one time MAC is naturally very fast but I'm not sure how
large key storages can be added for micro controllers: users should
probably use OTF-encrypted HDDs to protect key data and avoid wear
levelling issues of flash memory.

Anyway, I pushed out 0.5.4. of TFC out yesterday. Lot's of fixes for
stability and usability, signed installer that checks SHA512 hashes of
other files. Probably the most important feature is hiding 'when' and
'how much' communication takes place. This is done by sending a constant
stream of noise messages and commands from the transmitter unit the
receivers transparently discard. This exhausts OTP keyfiles very quickly
so I'd recommend using the CEV version.

-maqp


On 31.03.2015 05:49, Juan wrote:
> On Sun, 29 Mar 2015 00:46:08 +0200
> Markus Ottela <oottela@cs.helsinki.fi> wrote:
> 
>> A microcontroller as TCB doing OTP with HWRNG-generated keys. Sounds a
>> lot like the OTP-version of Tinfoil Chat ( github.com/maqp/tfc ). 
> 
> 
> 	Splitting the application in two rx/tx physically isolated
> 	devices is clever...Although using two laptops or two
> 	raspberries seems a bit overkill?
> 
> 
> 	Now I'm wondering how easy it would be to hack a
> 	microcontroller through its serial link. Of course "a
> 	microcontroller" is horribly vague. For instance, what about a
> 	microcontroller that can't execute code from ram? 
> 
> 
> 
> J.
> 
> 
> 
>