See comments inline

 

 

From: Ace [mailto:[email protected]] On Behalf Of Somaraju Abhinav
Sent: 02 February 2017 03:48
To: Jim Schaad <[email protected]>;
[email protected]
Cc: 'ace' <[email protected]>
Subject: Re: [Ace] draft-somaraju-ace-multicast

 

Hi Jim,

thank you for the review and I apologise for the delayed response - I was on
sick leave due to a surgery. Please see comments inline from the authors. 

 

Why restriction on reading messages?  It is not like an external observer is
not going to be able to see the lights go on or off.
[AS] There are several situations where lights are not visible but
(multicast) network data is accessible. Moreover, sensors (e.g. presence
detectors) are continuously talking to actuators and controllers without
necessarily having a visible effect on the lights. For several customers
privacy is a very important concern and is almost a given. The statement
"anybody can listen to the traffic and tell when sensors detect presence in
a building without even being in the building" is a very difficult sell.
Having said that, it is true that simply encrypting the multicast traffic at
the application layer is only a prerequisite to provide the privacy needed
and additional work is required (e.g. generating random messages at
different times). In that sense the symmetric solution is probably not much
better than the asymmetric solution. But the demand for privacy from
customers is very clear and the perception among them is that unencrypted
data implies poor security.

[JLS] I am sensing a problem here.  You have stated that there is a
requirement that encryption is a requirement that people are going to say
must be me.  However, below you have stated that if authentication is a
requirement then encryption suddenly becomes a non-requirement?  You appear
to be stating that there are circumstances where it is fine not to have the
data encrypted if one needs to know where it came from.

 

Consider the following case   I have a sensor in a room.  When the sensor
sees movement, it broadcasts a lights one command.  The command is picked up
by both the lightbulbs and by the security system.  The security system must
know which sensor provided the command and therefore no encryption is going
be needed here?  That just seems wrong.

 

Additionally, the situation where things are "continuously" talking would
seem to be a good place where one would want to install a controller and not
have the sensor directly talking to the actuator.  You don't want to flood
the actuators with trying to constantly turn on the lights.  Also the use of
actuators in this sense makes one think that this is a solution for things
other than lighting systems which is what people are complaining about.

 


The solution in section 4 does not seem to meet the following requirement
"Only authorized members of the application group must be able to read and
process messages."
[AS] You are right, we cannot satisfy the privacy requirement in Section 4.
We could extend the current solution to include a group wide encryption key
to meet this requirement. However, this will add additional latency to the
asymmetric solution.



This document needs to have a solution for dealing with nonce space
allocation for the cases where more than one sender is going be able to use
the same key.  This is going to be part of the problems with replay
detection as well as security considerations.

[AS] Okay. Will add some text in the next version of the draft for better
clarification. The idea as written in 4.3 (Nonce value) is to use the Client
ID along with the sender's sequence number to create the complete nonce for
replay and CCM processing.


Should the algorithms be using high water detection of sequence numbers
rather than the case of not yet used?  Or is that an application specific
type thing?

[SK] This is tricky since it can create all kind of new issues. One way to
handle if the sequence number of a sender is about to roll over is that the
sender requests a new key issued for the group by the KDC. Tricky part is if
there are multiple senders who are not reaching the roll over of their
sequence number then have to be forced to use a new key or there needs to be
some overlap between the old key and new key before every sender in the
group starts using the new key.

[JLS] Lots of spinning in graves from the idea of having a sequence number
roll over given the harsh requirements that a nonce (built from the sequence
number) must never be re-used twice for many of the algorithms that are
going to be used here.

 

I do not think that the current security requirements is sufficiently
strident to reflect both the threat of breakage, cross-breakage and
restrictions on where it should be used to pass muster.

[AS] I thing this will be the main discussion item in the webex. We will
make a proposal for the security guidelines section after the interim webex.

[JLS] A proposal before the call is better because then we have a starting
point for discussions as well as allowing people who will not make the call
be able to have some initial input on where discussions points should be
directed.







_______________________________________________
Ace mailing list
[email protected] <mailto:[email protected]> 
 <https://www.ietf.org/mailman/listinfo/ace>
https://www.ietf.org/mailman/listinfo/ace

________________________________________________________ The contents of
this e-mail and any attachments are confidential to the intended recipient.
They may not be disclosed to or used by or copied in any way by anyone other
than the intended recipient. If this e-mail is received in error, please
immediately notify the sender and delete the e-mail and attached documents.
Please note that neither the sender nor the sender's company accept any
responsibility for viruses and it is your responsibility to scan or
otherwise check this e-mail and any attachments. 

_______________________________________________
Ace mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/ace

Reply via email to