[TriLUG] interoperability: 802.11g and n

Joseph Mack NA3T jmack at wm7d.net
Sun Oct 4 14:20:04 EDT 2009


>> There's a decent article right on Wikipedia...

This, and a list of references to the mesh routing demons, 
set me on the right track.

My original interest was if the new dd-wrt running on the 
WRT54G-TM (with 32M memory) could do mesh routing. It looks 
like it can.

In the meantime I got a whole lot of other stuff straight.

802.11n is not mesh routing (as I thought). It's 100Mbps 
using MIMO (multiple in, multiple out) antennas, ie 2 or 
more antennas at each end. The multiple antennas are used to 
reduce interference, through among other things, on the fly 
phasing for each packet. Unless a large number of elements 
is in use, this seems to be a computationally intense 
procedure for not much gain to my mind. I didn't find the 
packet protocol underneath. I didn't find any examples of 
devices using it. I don't know if it's designed for trunking 
or for leaf nodes or if it's node agnostic. AFAIK it's not 
here yet.

802.11s is mesh routing, like used in the OLPC. The routing 
is described independantly of the packet protocol. 
Presumably a node could use 802.11a,b,g for the packets, and 
802.11s for the routing. I don't know if an OLPC can join an 
802.11g network out of the box or not. It seems in principle 
it could.

Most interest in mesh routing is for VOIP and in particular 
the military is interested in it for battlefield operations.
People like you and me don't rate in the setting the pace 
and direction of 802.11s.

>From what I can see, the routing used for current 802.11g 
links is just the standard wired ethernet routing with a 
default gw. In 802.11s, nodes (or leaf computers) have a 
view of the network much like that of a wired router running 
an internet routing protocol - no default gw and a large 
routing table.

There's a couple of differences between wired internet 
routing and 802.11s

o all hops aren't equal - there is reliability factor 
(weighting) for each hop

o packets arrive and leave via the same interface.

o routing is proactive - the nodes are looking for and 
updating routes continually even if there is no traffic 
asking for those routes. This means each node has to hold a 
table of all the routes and a large amount of traffic is 
involved in updating the routes. A wired network is 
reactive, nodes don't search for a route till traffic 
arrives for that route. How discovery works and updates are 
propagated in 802.11s I didn't discover - ie what is the 
equivalent of icmp. At least one 802.11s protocol uses UDP 
packets to a low port for some part of this. The large 
amount of traffic for route updates means that nodes that 
are asleep most of the time (eg sensors which wake up once 
an hour to send a reading) can't be 802.11s. I'm not sure 
how a non-s leaf node fits in (joins, leaves) to a 802.11s 
network.

Joe
-- 
Joseph Mack NA3T EME(B,D), FM05lw North Carolina
jmack (at) wm7d (dot) net - azimuthal equidistant map
generator at http://www.wm7d.net/azproj.shtml
Homepage http://www.austintek.com/ It's GNU/Linux!



More information about the TriLUG mailing list