Re: [lamps] Multiple drafts with PQ algorithm key encodings that are not compatible.

[Mike Ounsworth <[e-mail]>]

So, taking the CRYSTALS keygens as examples:

Dilithium:
01: \zeta <- {0,1}^256
Then to my eyes, the rest of the keygen is just hashing and some not horrible arithmetic.

Kyber:
01: d <- {byte}^32
Then to my eyes, the rest of the keygen is just hashing and some not horrible arithmetic.


What are the performance numbers for Round 3 keygens? My go-to was https://www.safecrypto.eu/pqclounge/, but it does not appear to have round 3, and I admit that I am having troubles navigating https://bench.cr.yp.to/primitives-sign.html to find keygen numbers.


I can certainly see why it would be attractive to only store / transmit the 32 byte seed. Are there any use cases that *can* tolerate the bandwidth, but *can not* tolerate re-running the keygen on demand?

Also, I assume similar arguments apply to SPHINCS+, XMSS, LMS, but not to Falcon since the Falcon seems to have heavier field computations, and to need more than a single random value (but I’m just reading the spec).


---
Mike Ounsworth
Software Security Architect, Entrust

From: Spasm <[e-mail]> On Behalf Of Blumenthal, Uri - 0553 - MITLL
Sent: September 29, 2022 12:00 PM
To: Tim Hollebeek <tim.hollebeek=[e-mail]>; Russ Housley <[e-mail]>
Cc: LAMPS <[e-mail]>
Subject: [EXTERNAL] Re: [lamps] Multiple drafts with PQ algorithm key encodings that are not compatible.

WARNING: This email originated outside of Entrust.
DO NOT CLICK links or attachments unless you trust the sender and know the content is safe.
________________________________________
Yes, it’s a tradeoff. Given the cost of the constraints, and the fact that computing power tends to consistently grow (without necessarily consuming much more power), but the bandwidth of the radio communications does not (e.g., there’s no way to squeeze significantly more bits through an HF “connection”), I strongly support favoring smaller structures to go across the wire, rather than larger structures that take less CPU cycles to process.

-- 
V/R,
Uri


From: Spasm <mailto:[e-mail]> on behalf of Tim Hollebeek <mailto:tim.hollebeek=[e-mail]>
Date: Thursday, September 29, 2022 at 11:47
To: Russ Housley <mailto:[e-mail]>
Cc: LAMPS <mailto:[e-mail]>
Subject: Re: [lamps] Multiple drafts with PQ algorithm key encodings that are not compatible.

Yes, I get that part, I just wanted to make sure I correctly understood what aspect of uni-qsckeys you were referring to.

It is a little unfortunate that both encodings are important for different use cases, and we can’t just pick one.  Perhaps different algorithm OIDs would be cleaner than fooling around with CHOICE and OPTIONAL all over the place.  But that’s probably going to get tricky too…

-Tim

From: Russ Housley <mailto:[e-mail]> 
Sent: Thursday, September 29, 2022 11:14 AM
To: Tim Hollebeek <mailto:[e-mail]>
Cc: LAMPS <mailto:[e-mail]>
Subject: Re: [lamps] Multiple drafts with PQ algorithm key encodings that are not compatible.

Tim:

There is a tradeoff between size and computation.  By including values that can be calcualted, the key gets bigger, but it is easier to consume.  However, more battery is needed to transmit and receive the bigger structure.

Russ


On Sep 29, 2022, at 10:28 AM, Tim Hollebeek <mailto:tim.hollebeek=[e-mail]> wrote:

Are you referring to the option for partial vs full encoding for some of the private keys, or something else?
 
-Tim
 
From: Spasm <mailto:[e-mail]> On Behalf Of Russ Housley
Sent: Thursday, September 29, 2022 10:12 AM
To: LAMPS <mailto:[e-mail]>
Subject: Re: [lamps] Multiple drafts with PQ algorithm key encodings that are not compatible.
 
Trimming the recipient list because the other messages needed moderation for "too many recipients"
 
draft-uni-qsckeys-01 makes the argument that for the private keys, one size does not fit every use case.  If there is consensus on this view, thenOPTIONAL fields or a CHOICE within the private key structure seems like a simple way forward.
 
I was really hoping that the public key would _always_ be an OCTET STRING.  No optional parts. 
 
Russ
 

On Sep 28, 2022, at 2:33 PM, John Gray <mailto:John.Gray=[e-mail]> wrote:
 
We are doing some interoperability testing with different vendors using the new Dilithium, Falcon and SPHINCS+ algorithms.   We have come across at least two drafts which are trying to specify the ASN.1 encoding formats for these algorithms.   However, the encoding formats are not compatible with each other.   I imagine the authors of these drafts should get together and come up with a common format (I have copied them on this email).   This means we must choose one or the other, or even worse, support multiple formats (which can lead to bugs).   Initially I started more than a year ago using my own encoding format for internal prototyping, but now need to interoperate with others outside of our organization, so a common format is definitely needed at this point.   😊
 
We fully realize the OID values will be changing once official OIDs are registered, (changing those are trivial), but the ASN.1 formats of the public and private keys is kind of important as well… 😊
 
For example:
 
The LAMPS group has a specification of Dilithium public keys in this draft:
https://datatracker.ietf.org/doc/draft-massimo-lamps-pq-sig-certificates/
 
the public key format is this:
 
The Dilithium public key MUST be encoded using the ASN.1 type
   DilithiumPublicKey:
 
     DilithiumPublicKey ::= OCTET STRING
 
The private key format is this:
 
     DilithiumPrivateKey ::= SEQUENCE {
         rho         BIT STRING,         - nonce/seed
         K           BIT STRING,         - key/seed
         tr          BIT STRING,         - PRF bytes (CRH in spec.)
         s1          BIT STRING,         - vector l
         s2          BIT STRING,         - vector k
         t0          BIT STRING,         - encoded vector
         PublicKey   IMPLICIT DilithiumPublicKey OPTIONAL
     }
 
 
In this draft:
https://datatracker.ietf.org/doc/draft-uni-qsckeys/01/
 
Dilithium keys have this encoding:
 
 
   DilithiumPublicKey ::= SEQUENCE {
       rho         OCTET STRING,
       t1          OCTET STRING
   }
 
 
  DilithiumPrivateKey ::= SEQUENCE {
       version     INTEGER {v0(0)}     -- version (round 3)
       nonce       BIT STRING,         -- rho
       key         BIT STRING,         -- key/seed/D
       tr          BIT STRING,         -- PRF bytes (CRH in spec)
       s1          BIT STRING,         -- vector(L)
       s2          BIT STRING,         -- vector(K)
       t0          BIT STRING,
       publicKey  [0] IMPLICIT DilithiumPublicKey OPTIONAL
                                       -- see next section
   }
 
The draft-uni-qsckeys does not cover SPHINCS+,  it does cover Falcon, but I don’t know of another draft that specifies Falcon.
 
There are also encodings for Kyber mentioned in two documents that I see.  There is an early   https://datatracker.ietf.org/doc/draft-ietf-lamps-kyber-certificates/  draft which mentions it is up to the document defining Kyber to give more details.    In the draft-uni-qsckeys draft it is more specific.    
 
https://datatracker.ietf.org/doc/draft-uni-qsckeys/01/
 
   KyberPrivateKey ::= SEQUENCE {
       version     INTEGER {v0(0)}   -- version (round 3)
       s           OCTET STRING,     -- sample s
       publicKey   [0] IMPLICIT KyberPublicKey OPTIONAL,
                                     -- see next section
       hpk         OCTET STRING      -- H(pk)
       nonce       OCTET STRING,     -- z
   }
 
Partial public key encoding:
 
KyberPrivateKey ::= SEQUENCE {
       version     INTEGER {v0(0)}   -- version (round 3)
       s           OCTET STRING,     -- EMPTY
       publicKey   [0] IMPLICIT KyberPublicKey OPTIONAL,
                                     -- see next section
       hpk         OCTET STRING      -- EMPTY
       nonce       OCTET STRING,     -- d
   }
 
Full public key encoding:
 
   KyberPublicKey ::= SEQUENCE {
       t           OCTET STRING,
       rho         OCTET STRING
   }
 
Is https://datatracker.ietf.org/doc/draft-uni-qsckeys/01/ meant to become the document that defines the key formats for all the PQ keys that will be standardized?    If not, then it should probably just refer to whatever documents will define the formats so that we can at least agree on one common format for the PQ keys.
 
 
Cheers,
 
John Gray
Entrust
Any email and files/attachments transmitted with it are confidential and are intended solely for the use of the individual or entity to whom they are addressed. If this message has been sent to you in error, you must not copy, distribute or disclose of the information it contains. Please notify Entrust immediately and delete the message from your system.
_______________________________________________
Spasm mailing list
mailto:[e-mail]
https://www.ietf.org/mailman/listinfo/spasm
 
_______________________________________________
Spasm mailing list
mailto:[e-mail]
https://www.ietf.org/mailman/listinfo/spasm