[Olsr-users] Mesh on Gondola

[Hesameddin Alizadeh]

Thank you Henning.
Here is some questions which is not clear for me:

1. TC messages are flooded to all other mesh nodes and OLSR build a complete topology graph base on all TCs  :--> This should take a long time which is depend on number of nodes on network to build a complete topology graph from all TCs.( is it memory of 32 seconds that you explained before ?!)
2. If the nodes are moving then seems OLSR could not build a complete topology graph because it is not fixed.
3. how OLSR decided to use minimal path cost without complete topology?


-----Original Message-----
From: Henning Rogge [mailto:(spam-protected)] 
Sent: Montag, 21. Dezember 2015 18:41
To: Hesameddin Alizadeh <(spam-protected)>
Cc: (spam-protected)
Subject: Re: [Olsr-users] Mesh on Gondola

On Mon, Dec 21, 2015 at 6:21 PM, Hesameddin Alizadeh <(spam-protected)> wrote:
> OLSRv1 keeps all directly and MPRs connection on his topology table and put the best route to the destination ( minimum packet loss ) in his routing table, am I right?

I think you missed a couple of steps.

OLSR (both v1 and v2) detect symmetric links to their one-hop neighbors with the HELLO message... by listening to their one-hop neighbors they learn about the symmetric two-hop neighbors.

All these links (both in the olsrd ETX extension and in OLSRv2 in
general) have a cost value (link metric attached).

The olsrd link metric is based on the multicast packet loss, the
olsrd2 metric is the DAT (directional airtime) metric (see https://tools.ietf.org/html/draft-ietf-manet-olsrv2-dat-metric-12).

olsrd/olsrd2 use (depending on configuration) the link metric to calculate an MPR set. The full set of one-hop neighbors is a valid MPR set.

The MPR set is announced in each HELLO.

OLSR (both v1 and v2) store which neighbor selected the local node as an MPR... these nodes are put into the TC messages.

The TC messages are flooded to all other mesh nodes.

OLSR builds a complete topology graph from all TCs.

OLSR uses a minimal path cost algorithm (usually Dijkstra) to calculate the shortest path to each destination.

The next hop of each shortest path will be used as the "next hop" of the routing table entry for the destination.

Henning Rogge