Re: [Ace] on signature verification times for sec192r1

[Hannes Tschofenig <[e-mail]>]

Hi Pascal,

there is no doubt that today's IoT hardware can run Internet security
protocols. This is why we worked on the DTLS profile in the DICE working
group, which has been finally published as an RFC a few days ago. Here
is the link:
https://tools.ietf.org/html/rfc7925

There are two separate issues that are being raised here in this
discussion, namely

* Low latency requirements (and Abhinav just sent a document around that
provides further details), and

* Security of the stored keying material.

During the meeting we had a discussion about the ability (or inability)
to use public key crypto on modern IoT hardware so that it hits the time
budget needed for the lighting domain.

I personally see the security of the stored keying material is an
orthogonal issue since additional security protection makes attacks
harder but of course not impossible.

Btw, when you mention prices of chips you also need to indicate what the
volume is. The price is very much dependent on the volume.

Furthermore, you mention some performance numbers below but they seem to
be incomplete since I don't really know what the constant
'nb_bloc_512bits' is. Could you please elaborate?

Ciao
Hannes

On 07/24/2016 11:50 AM, Pascal Urien wrote:
> I fully agree...
> 
> J3A081M  can be found at 10$ over the WEB
> 
> Futhermore this class of cheap device can process TLS or DTLS as
> illustrated in
> 
> https://tools.ietf.org/html/draft-urien-uta-tls-dtls-security-module-00
> 
> They could be used for numerous applications in the IoT
> 
> Rgs
> 
> Pascal
> 
> 
> 2016-07-23 23:59 GMT+02:00 Michael StJohns <[e-mail]
> <mailto:[e-mail]>>:
> 
>     On 7/23/2016 11:10 AM, Pascal Urien wrote:
>>     Hi All
>>
>>     J3A081M is a javacard device from NXP
>>
>>     The micocontroller should be the P5CD081V1A, which comprises a
>>     crypto processor
> 
>     There's a number of these from a number of vendors.  I'd actually
>     look at the A7xxx series of chips as they're designed to be
>     embeddable.  I've become a big fan of javacard style solutions over
>     the years.
> 
>     In any event, the number of relatively inexpensive public key crypto
>     accelerator chips (e.g. googl for "secure authentication chips") is
>     greater than zero and continues to climb.  And for not a lot of
>     money.  Estimating what from prices on Digikey, I'd think something
>     less than $.50 for Quantity large as of today and half that or less
>     in 1-2 years as its gets bundled into the "Swiss Army Knife" style
>     of process (e.g. support for wireless 900mhz plus ... plus ... plus
>     ... plus security...) (google for iot module secure element 900mhz
>     for example).
> 
>     Later, Mike
> 
> 
> 
>>
>>     The performances with the curve secp192r1 are the following (for
>>     ECDSA + SHA1)
>>
>>     Sign    = 40ms +  nb_bloc_512bits x 3.5 ms
>>     Verify  = 60ms  + nb_bloc_512bits x 3,5 ms
>>
>>
>>     By the way this chip has enough crypto ressouces for processing
>>     TLS or DTLS
>>
>>     Rgs
>>
>>     Pascal
>>
>>
>>     _______________________________________________
>>     Ace mailing list
>>     [e-mail] <mailto:[e-mail]>
>>     https://www.ietf.org/mailman/listinfo/ace
> 
> 
> 
>     _______________________________________________
>     Ace mailing list
>     [e-mail] <mailto:[e-mail]>
>     https://www.ietf.org/mailman/listinfo/ace
> 
> 
> 
> 
> _______________________________________________
> Ace mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/ace
>