[liberationtech] Mesh Networks?

[Charles N Wyble]

On 06/18/2011 08:46 PM, Danny O'Brien wrote:
>
> On Jun 17, 2011, at 2:27 PM, Steve Weis wrote:
> I'm not entirely adding new information here, but I spent a fair bit 
> of time researching WiFi mesh in 2004 when many of these technologies 
> were being first explored.

A lot has changed since then. Atheros chipset has won. OpenWRT has 
greatly matured. Dozens and dozens of production mesh networks exist 
(ranging from campus/town/enterprise deployments of 5 to 25 nodes up to 
several thousand nodes covering large urban metro areas).

> The general concern then was that building mesh on 802.11 commodity 
> hardware was  challenging because the niggling details of theoretical 
> mesh routing models were still in their infancy then, and applying 
> them to the real world (and a real RF environment) was exposing 
> serious and hard to debug challenges, especially on a platform that 
> wasn't built for mesh.

Yep. Linksys routers with broadcom chipset are horrible. However amazing 
things have been done with them. Check out http://ubnt.com for what 
folks are using now.

>
> I think when many people think of mesh now, they're still considering 
> using WiFi, but they may be unaware of the already large amount of 
> practical knowledge about the limitations of commodity WiFi tech to 
> pull this off.

Um. Define commodity WiFI? You mean routers from bestbuy/frys? If they 
have atheros chipsets and external antenna connectors, you can flash 
OpenWRT and have an amazing package. Stay away from broadcom chips. Or 
just get the packaged solutions from Ubiquity and call it a day. (well 
you still need to flash OpenWRT or the VT firmware but that's a trivial 
exercise).


>
> Effective mass hacking, I feel, comes from having a wide variety of 
> fairly well-known and commoditised tools that can be repurposed and 
> composed together to new ends. I think everyone wants that commodity 
> tool to be WiFi, but I think the last few years have shown that it's 
> by no means easy to do this.

It's trivial. It's known. It's mature. It's boring. Mesh is here to stay.

>
> Alternatively, we may have enough practical knowledge now to build 
> something stable and reproducible out of it, but the knowledge is 
> badly distributed among a lot of practictioners, some of whom may 
> already have given up, and some of whom may be trapped in attempting 
> to provide commercial services out of what they know.

Villagetelco.org is where a whole lot of us hang out. Come over and say 
hey. :)


>
> I'd love a frank summary from the Meraki, LocustWorld, PortlandTelco 
> and the 802.11s folks of what's been happening in the last few years. 
> Maybe this information is being collated somewhere: but I think it 
> would be more rewarding to collate that past experience, than start 
> from scratch all over again.

Mesh has arrived. Mesh has won. Ubiquity Networks + OpenWRT enables it 
all. We've just been moving along, quietly building out real networks. 
Everyone else talks about it, we just sat down and did it. Over and over 
and over. We build. We test. We tweak. We pass packets.

-- 

Charles N Wyble charles at knownelement.com @charlesnw on twitter
http://blog.knownelement.com
Building alternative,global scale,secure,
cost effective bit moving platform
for tomorrows alternate default free zone.