Mesh Link Establishment
draft-kelsey-intarea-mesh-link-establishment-06

I am sorry that I have only had time for a very brief review of this document. I should be happy if you are able to say that any of my Discuss points is rubbish based on my reading too fast.

-------

Section 4

   The purpose of the third, detecting neighboring
   devices, is to make link management more efficient by detecting
   unreliable links before any effort is spent configuring them.


Do you not think "detecting neighbor devices" is an odd name for a 
feature which is intended to detect unreliable links before any effort
is spent configuring them?

Similarly the title of section 4.3.

Isn't this link discovery? (A link not being an interface :-) For
example, suppose there are two links that reach the same neighbor?


But in section 8...
   Link configuration
   and advertisement messages MUST be sent with an IP Hop Limit of 255,
   either to a link-local unicast address or to the link-local all-nodes
   (FF02::1) or all-routers (FF02::2) multicast addresses.

...I believe that what you are defining is effectively a link state
protocol by full flooding. This is not a discovery protocol. I am very 
wary of such a protocol being defined (we have been there before and it
sucks!)

At the very least, this document needs to be clear whether it is a 
discovery protocol, a link parameter negotiation/discovery protocol, or
a link parameter flooding protocol. In the latter case, I would like the
document to explain the scope and purpose of this wholesale flooding.

On the other hand, part of the problem may be that you intend that MLE
should be able to send (from a central location) configuration
information that is to be used at each node (or some set of nodes) in 
the network. Section 11 gives this hint. Yet that is a very different 
use of the protocol and many features of that operation (such as how
thrashing between two senders is handled) are left for the reader to
ponder.

---

Section 5 needs to indicate how a parser knows the length of an MLE 
message. How do I know when to stop parsing TLVs and start parsing the
MIC (in the 0 case) or stop parsing the message (in the 255 case).

---

Section 7

   With the exceptions of the Source Address TLV and Parameter TLV, an
   MLE message MUST NOT contain two or more TLVs of the same type. 

You need to say how such errors are handled.

However, I think you have hamstrung yourself for the future. Why not add
a column to the registry that says "Multi-occurrence allowed" and then
state that RFCs defining new TLVs must state whether multiple occurrences
are allowed.

---

Section 7.1

   The Source Address TLV (TLV Type 0) has a Value containing a byte
   string representing a link-layer address assigned to the source of
   the message.

Does "a byte string representation" have a specific meaning in the 
context of IEEE 802.15.4? To me it is not specific enough to write code.

---

Section 7.6

"encoded as an N-bit unsigned integer."

This is curious as the TLV length can only indicate full octet lengths.

Is N allowed to be other than an integer multip of 8?

---

Section 7.7

   Size                The size in bytes of the included neighbor link-
                       layer addresses, minus 1.  This supports
                       addresses of lengths 1 to 16 bytes.

This is ambiguous. Do you mean the total size of all addresses, or the 
size of a single address with the requirement that all addresses must be
the same size?

Yet later in the same description of the data format...

   Address             A link-layer address of a neighbor.

... which says that only one address is present (and misnames the field
compared to the figure).

---

Section 7.8

   Value               A byte string containing the new value of the
                       parameter.  The format of this value is
                       determined by the particular parameter

And yet the formats don't appear to be described when the parameters 
are listed later in the section.

---

7.9.  MLE Frame Counter

   The MLE Frame Counter TLV (TLV Type 8) has a Value containing the
   sender's current outgoing MLE Frame Counter, encoded as an 32-bit
   unsigned integer.

What is an MLE frame? Is this a count of messages or L2 frames or...?
How does this wrap?
When is it reset?
Is the 32 bit integer in line format?

Section 5 gives some information about this with cross-pointers to [CCM]
Section 9 tells us how to handle received frame counters.
But 7.9 needs to point to these other sections, and Section 8 should
explicitly state about setting the counter.

I have no objection to the publication of this document, but there are some points that need work...

1. In section 6, it is unclear as to how Command Type 2 differs from a combination of Command Types 0 and 1.  In what instances would a sender use Command Type 2?  It seems to me that this *could* be an exchange like: 1) A sends CT==0 to B, 2) B sends CT==2 to A, and 3) A sends CT==1 to B.  Is that the expected usage?  If so, it should be explained.  I was expecting to see Section 10 do that, but that didn't materialize.

2. Some of the specified Command Types and TLVs are obviously link-local in nature.  Are there Command Types and TLVs that are only network-wide?  I would like to see some description of how far each of the Types and TLVs should be propagated.

3. This is related to Barry's comment about destination address selection.  Section 8 should tighten up the guidance on address selection.  This should be tied to the point raised in #2 above, in that the destination address needs to take the scope of the information contained in the message into account.

4. It is unclear to me what Section 9 is saying about "reserved Command Types" and "reserved TLV Types".  Are those just values that are not recognized?

The draft seems okay, although I'd like to understand why there isn't a stronger requirement around replay protection or see that fixed with stronger language.  The Security Considerations section says:

   Because of this, implementers must
   be careful in how they use information obtained from these possibly-
   replayed messages.  For example, information from unsecured messages
   should not be used to modify any stored information obtained from
   secured messages.

and section 7.4 provides a way to do this with a challenge:

   The recommendations in [RFC4086] apply with
   regard to generation of the challenge value.  The byte string MUST be
   at least 4 bytes in length and a new value MUST be chosen for each
   Challenge TLV transmitted.  An important part of replay protection is
   determining if a newly-heard neighbor is actually present or is a set
   of recorded messages.  This is done by sending a random challenge
   value to the neighbor and then receiving that same value in a
   Response TLV sent by the neighbor.

I'm not familiar with 802.15.4 security sorry, so you
may need to educate me a bit, and I had to review this
in a hurry so might be wrong below (and sorry again:-)
but I had the following questions to discuss:

(1) I don't get why its ok that there is no key
management defined, either here or somewhere else via
reference. That could be fixed via some more text in
section 3 perhaps, if there are other specs that do
define how to do key management, e.g. to reference one
(or more) that do define that and to say that other
uses of this need to do similarly. Absent that, maybe
we're back to looking at BCP107, but I'm not sure we
want to (or hopefully, need to:-) go there.

(2) The overall security model is a bit complicated, I
wasn't clear whether the combinations of security
being on and off (i.e. initial byte = 0 or 255) adds
up to a good outcome. There are also cases (e.g.
joining) where where MLE security MUST NOT be used but
link layer MUST be used.  Has someone done an analysis
of how that all stacks up together? Can you point me
at that? (Sorry for the fairly vague question.)

(3) As a particular case of (2), I don't get how the
device joining without any keys thing is meant to
work, nor under exactly what conditions. Don't you
need to specify this or at least state the
requirements for it to work if specified elsewhere?

(4) I'm not clear that the frame counter + key cannot
repeat, esp. since we don't have the key management
detail here and that there seems to be a way for one
node to tell another which frame counter value to
start from (is that right or is that just telling a
new peer what my current frame counter is at? If the
latter, why not just get that by observation?). I do
see you say that they MUST NOT be repeated, but if
there are shared keys over >1 device, or, if its
allowed to not use the same frame counter for link and
MLE but is allowed to use the same key, then that may
not be implementable.