[Olsr-dev] MPR selection algorithm in olsr

[Teco Boot]

Op 25 jul. 2012, om 18:08 heeft Henning Rogge het volgende geschreven:

> The main problem with the current code is that it cannot handle
> situations where you have multiple links between two nodes.
> 
> Its not even a fault of the MPR algorithm, its a fault of the
> Neighbor-2 database, which is Node (and not link) specific. OLSRv1 has
> no database to store the specifics of 2-hop neighbors, because it
> wasn't necessary for the original protocol (which had no link metric
> at all).

This may is true for topology reduction. For flooding, I have doubts.
See also IETF MANET ML archive for details, thread "MPR selection per interface - race conditions".

Teco


> 
> Henning
> 
> On Wed, Jul 25, 2012 at 6:05 PM, Bidur Devkota <(spam-protected)> wrote:
>> I am thankful to Teco Boot for those links which were of great help for me.
>> 
>> Further, I am looking if someone could tell about the location of the
>> function in the OLSR Code which runs the MPR selection algorithm.
>> I could see something in "src/mpr.c", since this is my first time parsing
>> the source code, I would be more confident with some more advice from
>> experienced person. Any resources or tips for beginners which will guide for
>> understanding and changing code will be of great value for me.
>> 
>> Thanks.
>> 
>> 
>> On Fri, Jul 20, 2012 at 5:25 PM, Teco Boot <(spam-protected)> wrote:
>>> 
>>> olsr.org MPR selection is a topic that needs further study.
>>> 
>>> Few years back, there was some R&D effort to enhance OLSR.
>>> http://wwwhome.math.utwente.nl/~haanr/Data/paper_WRECOM07.pdf
>>> http://doc.utwente.nl/65399/1/memo1892.pdf
>>> There is an quite old branch, with link costs, using data rate probing
>>> plugin. I think it has an updated MPR selection also (to be checked).
>>> http://sourceforge.net/projects/olsr-lc/
>>> 
>>> You could check the master branch also, and implementation of
>>> MPR_COVERAGE.
>>> 
>>> It is an open question how much effort we put in current olsr (version 1),
>>> with ETX/ETT.
>>> I welcome an analyze of the problems, if any, and why MPR flooding is
>>> turned off.
>>> You will see there is some confusion between MPR redundancy and TC
>>> redundancy, maybe even in the code....
>>> You may easily get confused reading olsrv2 draft also, where there are
>>> Flooding MPRs and Routing MPRs.
>>> 
>>> Teco
>>> 
>>> Op 20 jul. 2012, om 10:36 heeft Bidur Devkota het volgende geschreven:
>>> 
>>>> Hi All,
>>>> I have gone through a couple of resources for OLSR MPR selection. As
>>>> default method, OLSR protocol uses Greedy method for the purpose of
>>>> Multipoint Relay (MPR) selection. Also, different optimization have been
>>>> proposed for the MPR selection based on node density, necessity first, etc.
>>>> 1. I am seeking to know about the best way to optimize it. Any direction
>>>> in this regard is appreciated.
>>>> 2. Also, the RFC for OLSR (http://www.ietf.org/rfc/rfc3626.txt) shows
>>>> just the 2003 version. How can I know about the CHANGES/UPDATES done in MPR
>>>> selection in the NEWER version of OLSR.
>>>> 
>>>> Thanks !
>>>> 
>>>> Bidur
>>>> --
>>>> Olsr-dev mailing list
>>>> (spam-protected)
>>>> https://lists.olsr.org/mailman/listinfo/olsr-dev
>>> 
>> 
>> 
>> 
>> --
>> Regards,
>> Bidur Devkota
>> Asian Institute of Technology, Thailand
>> 
>> 
>> 
>> 
>> --
>> Olsr-dev mailing list
>> (spam-protected)
>> https://lists.olsr.org/mailman/listinfo/olsr-dev
> 
> 
> 
> -- 
> Steven Hawkings about cosmic inflation: "An increase of billions of
> billions of percent in a tiny fraction of a second. Of course, that
> was before the present government."