[aodvv2-discuss] Re: Questions - Sequence numbers
- From: Charlie Perkins <charles.perkins@xxxxxxxxxxxxx>
- To: aodvv2-discuss@xxxxxxxxxxxxx
- Date: Thu, 16 Apr 2015 09:46:53 -0700
Hello folks,
More follow-up below...
On 4/16/2015 6:58 AM, Victoria Mercieca wrote:
This email is about Sequence numbers (Section 6.4 in Version 9a or
Section 4.4 in Version 9b):
* The draft says we can use multiple SeqNums ("one per IP address").
What if there are multiple IP addresses on each interface - should
they have a SeqNum each, or share one SeqNum? Should it be one
SeqNum per interface? Or since we associate the SeqNum with a
route to a Router Client, should it be one SeqNum per Router Client?
You can use as many SeqNums as you like, up to one per IP address.
I would *prefer* to specify that there is one SeqNum per *node*.
However, some people
go ballistic when they see anything about nodes, and it does not hurt to
have SeqNums
associated with addresses.
* If using multiple SeqNums, should we store the sequence number in
the interfaces list? (I have said we could do this in the draft)
I am not sure what this means. The association of SeqNum is to IP
address, not interface.
* When sending a RREQ on multiple interfaces, would the sequence
number be the same in all of them?
If it is the same RREQ, yes the SeqNum would be the same.
* When sending retries for a RREQ, is the SeqNum updated?
It will work either way, and so does not have to be specified. If you
want to
be definite about it, then you should pick the way that forces more traffic
but enables more routes to be discovered -- namely, updating upon each
retry.
Various decorations to AODVv2 enable more efficient utilization of the
SeqNum space. To be brief, updating the SeqNum partially obsoletes
the routing information in all AODVv2 routers that do not see the update.
This bit of obsolescence isn't very harsh, but it does (for instance) mean
that iRREP would work less often.
* Is TargSeqNum ever used in the RREQ? Is it something to do with
iRREP? If so should we have references to it in the main text?
We could move the specification for including TargSeqNum in RREQ into
the iRREP
specification, but I don't think it hurts to keep it in the main
specification. Either way...
Regards,
Charlie P.
Other related posts:
