Re: [hiprg] draft-irtf-hiprg-rfid-00.txt

[Robert Moskowitz <[e-mail]>]

One more point on 'signatures', see below...

On 07/06/2010 02:18 PM, Robert Moskowitz wrote:
> In sec 1.5 you have "imitator", do you mean "initiator"?
>
> In sec 2.3 you have "nul".  This is the only place with this spelling.
>
> In sec 2.3:
>
> each Hmac beeing assocaited to a particular key.
>
> should be
>
> each Hmac being associated to a particular key.
>
> I have problems with the HIP parameter named  "Signature-T".  This is 
> a MAC and should be labeled accordingly.  The lable "Signature" should 
> be reserved for digital signiture algorithms like RSA, DSA, and ECDSA.

Signature usually has the security attribute of non-repudiation.  This 
is because only the owner of the private key can create it.  MACs can be 
created by any owner of the shared secret.  When there are only two 
parties owning the key as here, each party has the level of assurance as 
to where the packet originated at.  But in protocols like 802.1AE and of 
course 802.11 group key, there is no such assurance as any owner of the 
key can spoof packets from any of the other owners.  This lack of 
non-repudiation attribute is why this parameter should NOT be called 
Signature-T.

>
> You use KI and ki seemingly to mean two different things.  You should 
> separate your labeling to avoid potential confusion.  (unless I am 
> confused and they ARE the same thing and just they should be the same 
> case).
>
> It looks like your Initiator HIT is a 128 bit random value.  I feel 
> this is a BAD THING, as HITs in BEX and DEX are used as IPv6 addresses 
> in applications and implementors might think that BEX-T HITs can be 
> used thusly as well.  This MAY produce undesired routing behaviour if 
> the 'prefix' is that of a routable IPv6 address.  You can do it one of 
> two ways.  Use the same /28 prefix as BEX and ask for a HIT Suite ID 
> or get a different /28 prefix.  The first option limits your random 
> value to 96 bits and eats up a HIT Suite.  The later requires you to 
> get your own allocation.
>
> I do not see what use the HIP_TRANSFORM is put to.  Your MACing is 
> SHA-1 and there is no encryption within BEX-T.
>
> Speaking about SHA-1, in HIP-bis, we are moving away from a fixed 
> hash.  What is th3e current state of hash function implementation in 
> RFID devices?
>
> Finally what is the source of random numbers in RFID devices?  Do they 
> have some hardware random generator, do the use a PRF with a seed from 
> some source (and what is it), or some other approach?
>
> Final thought before I hit the send key....  Consider making the name 
> of this "the HIP Tag EXchange" or HIP TEX. :)
>
>
>
>
> _______________________________________________
> hiprg mailing list
> [e-mail]
> https://www.irtf.org/mailman/listinfo/hiprg
>