RE: [manet] AODV Question

[DANIEL BYRNE <daniel.byrne@adtran.com>]

>I guess the point is not whether you would like to or not, but rather if
>it is feasible with 255+ hops. And we did talk about AODV...
Yes we did, but everyone here is now saying that AODV doesn't work well over 50 or more hops.  

>In the Internet I guess that an average hop count would be 20-30 hops
>(with max around 60, correct me if I am wrong).

So the Internet has 20-30 max hops right now.  So what.  The internet backbone is composed of very long expensive cables that someone has to pay for.  The reason I started this thread was because that I was considering whether or not a radio multihop network could replace expensive long haul lines between routers.

The theory is if you could replace long, expensive, fiber optic cable runs with wireless relay points that could double as routers, then potentially anyone could own and operate a long haul backbone.  Obviously this is a topic for another thread.  However, to divulge, if it could be made cheaper to pop down 200 wireless nodes in a chain configuration rather then lay fiber I believe there could be a large market for chain topology wireless networks.  In which case I guess AODV would either have to be modified, replaced with a more capable routing protocol or something else yet undiscussed.  I'm of course just theorizing right now.

As far as bandwith is concerned, I know of one current product that delivers DS3 equivalent bandwidth over 25 miles.  That is enough bandiwth to make 768 residential ADSL customers pretty happy.  255 of these in a row would give you a 6375 mile long broadband pipe, all without wires.  That's long enough to extend from Juneau Alaska to Tampa Bay Forida.  The extra hops would be used to service individual communities, or various corporate offices along the way.  All with leasing a single DS3 line from  Ma Bell.

The only major stumbilng block I see in such a configuration is node latency.  However, I don't think the AdHoc routing protocols should concern themselves with the speed of the devices that they run themselves on.  We'd let the hardware designers figure that one out later.

_______________________________________________
manet mailing list
manet@ietf.org
https://www1.ietf.org/mailman/listinfo/manet