<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
<HTML><HEAD>
<META http-equiv=Content-Type content="text/html; charset=iso-8859-1">
<META content="MSHTML 6.00.6001.18063" name=GENERATOR>
<STYLE></STYLE>
</HEAD>
<BODY bgColor=#ffffff>
<DIV><FONT face=Arial size=2>Hello,</FONT></DIV>
<DIV><FONT face=Arial size=2></FONT> </DIV>
<DIV><FONT face=Arial size=2>I'm currently testing olsrd with some small devices 
(iPaq, eeePC, HTC...) and I've read a lot of things about the configuration 
file. Most of the time, it's optimized for big networks with a lot of statics 
nodes.</FONT></DIV>
<DIV><FONT face=Arial size=2></FONT> </DIV>
<DIV><FONT face=Arial size=2>I would like to optimize the configuration for a 
small mobile network (i.e. 10 nodes) with the following 
<DIV id=result_box dir=ltr>characteristics:</DIV>
<DIV dir=ltr>- very mobile nodes and no static node, no ap</DIV>
<DIV dir=ltr>-olsrd running on small devices (pda and htc with headset, 
etc.)</DIV>
<DIV dir=ltr>-no internet gateway</DIV>
<DIV dir=ltr>-no need to interact with non-olsr neighbours</DIV>
<DIV dir=ltr>-the network have to support voip traffic but users will speak very 
few</DIV>
<DIV dir=ltr>-the network must be very reactive and support small voip traffic 
(or small P2P communications)</DIV>
<DIV dir=ltr> </DIV>
<DIV dir=ltr>In fact, to imagine, just think about small teams of firemen, about 
people in charge of security during outdoors events, etc.</DIV>
<DIV dir=ltr> </DIV>
<DIV dir=ltr>I've read the archives and some documents speaking about olsrd and 
noticed some advices/tips/rules/etc and finally filled my configuration file 
like that:</DIV><SPAN lang=FR>
<DIV dir=ltr>
<P>DebugLevel 0 #The daemon runs in the background - Users don't need it...</P>
<P>IpVersion 4</P>
<P>FIBMetric "flat"</P>
<P>AllowNoInt no # We are not in a PCMCIA/USB hotswap environment and the 
interfaces should be always avaible...</P>
<P>Willingness 4 # Dynammically based on battery/power status seems to be a good 
idea for small devices. What about 7 ???</P>
<P>IpcConnect {</P>
<P>MaxConnections 0 # No GUI front-end in the wince, and no need to 
connect to the daemon</P>
<P>}</P>
<P>UseHysteresis no # delays neighbor registration... Not a good idea for a very 
reactive network...</P>
<P>Pollrate 0.05</P>
<P>NicChgsPollInt 60.00 # Interface configuration should never change... Is it 
possible to disable this checking ?</P>
<P>TcRedundancy 2</P>
<P>MprCoverage 1 # Defaults to 1 , and any other setting will severly reduce the 
optimization introduced by the MPR secheme says the documentation. But I often 
see values like 5, 7... So ?</P>
<P>LinkQualityLevel 2</P>
<P>LinkQualityFishEye 1</P>
<P>LinkQualityWinSize 100</P>
<P>ClearScreen no</P>
<P># NO PLUGINS</P>
<P>Interface "IFXX" {</P>
<P>AutoDetectChanges: no # turn off to save CPU</P>
<P></P>
<P>HelloInterval 2.00</P>
<P>HelloValidityTime 6.00 # This value must be larger than than the HELLO 
generation interval to make any sense.</P>
<P>TcInterval 0.50 # TcInterval 1.0 seems to be a bit too high 
together with the fish-eye, because this means that all nodes farther than 3 
hops away do only receive a TC packet every 39 (13*3) seconds (assuming the 
broadcasts are not lost in between). Better is probably TcInterval 0.5 as said 
in the README-Link-Quality-Fish-Eye.txt. And TcInterval should be shorter than 
the HelloInterval. This may overcome the "routing loop problem" encountered 
occasionally because topology info is spreaded more frequently.</P>
<P>TcValidityTime 1.50 # This value must be larger than than the TC generation 
interval to make any sense.</P>
<P>MidInterval 10.00</P>
<P>MidValidityTime 30.00 # This value must be larger than than the MID 
generation interval to make any sense. It appears that we 
should "consider to use MidValidityTime >= TcValidityTime (Since 
MID(multiple-identity)-info is used to parse traffic-control messages, itīs 
nescessary to track MID at least as long as TC-info. MID tells about nodes with 
multiple interfaces, a ka multiple IPīs meaning the same hop. since that info 
usually doesīt change "that often" ;-) , itīs quiet safer to extend 
MidValidityTime , like e.g. MidValidityTime = TcValidityTime +30. That way you 
avoid some quiet nasty effects, (learned it the hard way, too, in Berlinīs 
Freifunk-mesh)." AND "There is no reason to use smaller values for 
MidValidityTime than for TcValidityTime."</P>
<P>HnaInterval 5.00</P>
<P>HnaValidityTime 15.00 # This value must be larger than than the HNA 
generation interval to make any sense.</P>
<P>}</P>
<P>What do you think about it ? Could you give me some advices for this 
configuration or share your own experience ?</P>
<P>Thx a lot</P>
<P> </P></SPAN></DIV></FONT></DIV></BODY></HTML>