[Sis-dtn] SIS-DTN BPsec Profile

[sburleig.sb at gmail.com]

My sense of integrity vs authority, which may well be wildly wrong, is that
integrity can be provided by a checksum or CRC or by a signature computed in
a symmetric key that everybody knows, but authority can only be provided by
a signature computed in the sender's private key (verified in the sender's
known public key).  I strongly suspect it's not that simple, though.

 

Scott

 

From: SIS-DTN <sis-dtn-bounces at mailman.ccsds.org> On Behalf Of Dr. Keith L
Scott via SIS-DTN
Sent: Tuesday, November 22, 2022 11:04 AM
To: sis-dtn at mailman.ccsds.org; sea-sec at mailman.ccsds.org
Cc: Peter Shames <peter.shames at jpl.nasa.gov>
Subject: [Sis-dtn] SIS-DTN BPsec Profile

 

Greetings,

 

We have a joint meeting scheduled on Friday Dec 2.  This is nominally one of
the monthly meetings to discuss the new BPsec Green Book, but I'd like to
propose taking the December 2 meeting to discuss the BPsec Blue Book
Profile.

 

I had a discussion w/ Howie the other day that resulted in a number of
changes to the document:

 

Authenticity

Antonias had several comments around authenticity and whether or not it made
any sense to provide integrity without authenticity.  I could envision a
mission that wanted to provide data integrity on the science data it was
returning, but might not need/want to provide authenticity.  The assumption
here would be (I suppose) that it wouldn't make sense for anyone to fake the
data (e.g. a faked image purportedly from Pluto Express showing a sign on
the surface "I want to be a planet again."?)

 

That said, it seems like the services missions might want to choose from /
implement are:

*	Integrity
*	Authenticity
*	Confidentiality

 

[I'll admit to being a bit confused by this; MY model for authenticity would
be to use some sort of signed hash on the primary bundle block (which
includes the source EID), though I suppose other mechanisms are possible].

 

In the document I tried to use "authenticity / integrity" where appropriate,
and to otherwise mention authenticity where I thought it was appropriate.
I'd be interested if folks think I got close to right.

 

I still need to add some text around the 'pick-list' notion of integrity /
authenticity / confidentiality above.

 

Security Contexts

I added some text about security contexts and moved other text around so
that security contexts now show up earlier than they used to.

 

Default Security Contexts

RFC9173 contains a set of default security contexts for BPsec:

*	Integrity Security Context BIB-HMAC-SHA2
*	Security Context BCB-AES-GCM

 

I think the questions I'd like to get at at next week's telecon is:

*	Do we need a set of default security contexts for the CCSDS Profile
of BPsec?

*	I think so.  Maybe not even mandatory to implement but at least a
defined set that can be used for testing?

*	If the answer to the above is in fact 'yes' - what should we use for
the default profiles?  The current book has (I think) essentially RCC9172
pulled in, but then it looks like somebody (apologies, the changes are only
marked as 'Author') seems to have suggested changing some of the recommended
key sizes.

So, if we could at least start talking about a nominal set of security
contexts for the profile I think that would get us a LOT further down the
road to getting the book out.

 

                                v/r,

 

                                --keith

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.ccsds.org/pipermail/sis-dtn/attachments/20221122/d4b9cf8c/attachment-0001.htm>