ManchesterWireless

Mesh networks and why they don't scale

For a long time, companies promoting wireless "mesh networks" have ridden a wave of media hype, as pundits queue up to present ad-hoc wireless networks as the answer to the "last mile problem".

We weren't the first when we pointed out that mesh networks don't scale, but since then startups have continued to burn their way through venture capital, selling a network model which just won't work.

The issue hit the news again, when Francis daCosta, founder of mesh networking company meshdynamics, admitted the flaws inhernet in Wi-Fi meshing in his article 'The Ugly truth about mesh networks'.

The article has spawned plenty of responses, notably from Sascha Meinrath of the CUWiN project, and Jim Thompson, formerly of Wayport, Musenki, and Vivato. Jim's analysis is particularly telling, and he offers some interesting conclusions.

1) '802.11 won't scale. Mesh based networks based on 802.11's MAC won't scale. I agree.

2) Mesh can scale.

3) Community wireless networks will happen, and there is nothing that the carriers or telcos can do that will stop it.

The most interesting nugget to take from this article is that Linux drivers for the Atheros chipset allow you to write your own wireless MAC layer!

We've known for a long time that Wi-Fi is extremely ill-suited for what we want to use it for, but I had never realised that we can take advantage of commodity hardware to design and implement a Community Wireless MAC protocol.

Instead of implementing mesh routing protocols over a network that won't scale, why don't we design ourselves a MAC layer that will?

Posted by rob at July 8, 2004 02:33 PM