[Olsr-dev] Documentation on OLSR-ng

[Hannes Gredler]

hi hilton,

On Wed, May 20, 2009 at 07:21:00PM -0300, Hilton Garcia Fernandes wrote:
| i've just found out your very interesting presentation named "OLSR-NG 
| status-report", at www.olsr.org/~hannes/olsr-ng-v2.pdf
| 
| We're implementing a municipal wireless network and would like to know a 
| little more about the routing algorithms used in OLSR-ng. 

its a basic Dijkstra algorithm, albeit tuned for speed and efficiency.
with O(L)*lg(N) complexity (L being the number of Links and N the number of
nodes).

| May we think that MPR is no longer in use, and that BATMAN is another name for 
| OLSR-ng ?

MPR is still in the code, altough in most delpoyments like in vienna (400 routers)
or berlin (700 routers) it has been turned off (tc-redundancy = 2).

some history:
  back in 2005, some of the folks deploying olsrd had concerns about the scaling of
  link-state protocols - they decided to work on a entire new routing protocol (batman),

  the olsr-ng team was believing that fixing the existing olsrd implementation
  (which had a route calculation complexity of O^3.) was enough, and the recent
  deployments in berlin, leipzig and vienna proves that.

| In other words, could you please be as kind as to point us to some 
| documentation on OLSR-ng algorithms ?

nothing fancy - just correctly implemented dijkstra and RIB (evaluation).
all the magic is in olsr_spf.c, process_routes.c and routing_table.c

we have put special emphasis to carefully document the code. it should not
be that difficult to familiarize yourself. if you have questions, please
ask them on the developers mailing-list <(spam-protected)>.

HTH,

/hannes