[consume-routing] RE: New Not "mobile mesh" released under GPL (fwd)

Eugen Leitl eugen@leitl.org
Wed, 24 Apr 2002 12:39:54 +0200 (CEST)


---------- Forwarded message ----------
Date: Wed, 24 Apr 2002 01:00:52 +0100
From: Geoff Jukes <geoff@communitywireless.org>
Reply-To: consume-routing@lists.consume.net
To: 'R.D. Hammond' <muaddib@DATABIT7.COM>
Cc: 'Jon Anderson' <jon@LOCUST.CO.UK>, consume-routing@lists.consume.net,
     MOBILEMESH-LIST@lists.mitre.org
Subject: [consume-routing] RE: New Not "mobile mesh" released under GPL

Just a point:

"There are now at least 19 companies developing proprietary wireless
mesh routing technologies, all trying to replicate the original Internet
in a wireless form."

http://college4.nytimes.com/guests/articles/2002/03/04/905900.xml
http://www.corante.com/bottomline/
https://ssl.cpsr.org/pipermail/india-gii/2002-March/000531.html

So we (hopefully) are inventing the V8, not a Scirocco, or a VW, or a
merc, but an Engine. A Hydrogen powered one. Heck, maybe I'm a dreamer,
but the world needs dreams.

I can't speak for Jon, but I have a concept (and believe me it's basic)
but it is accurate (as a concept), and visual:

http://davis.wpi.edu/~matt/courses/soms/

Does that help?

Geoff
CommunityWireless.org

-----Original Message-----
From: owner-mobilemesh-list@lists.mitre.org
[mailto:owner-mobilemesh-list@lists.mitre.org] On Behalf Of R.D. Hammond
Sent: 24 April 2002 00:29
To: Geoff Jukes
Cc: 'R.D. Hammond'; 'Jon Anderson'; consume-routing@lists.consume.net;
MOBILEMESH-LIST@lists.mitre.org
Subject: RE: New Not "mobile mesh" released under GPL

Even before the Scirocco was produced it had a name, the "Type 39a".
This stops
confusion amoungst developers. Its not MM, its different. lets give it
its own
name.

rdh

On Wed, 24 Apr 2002, Geoff Jukes wrote:

> Date: Wed, 24 Apr 2002 00:05:05 +0100
> From: Geoff Jukes <geoff@communitywireless.org>
> To: 'R.D. Hammond' <muaddib@databit7.com>, 'Jon Anderson'
<jon@LOCUST.CO.UK>
> Cc: consume-routing@lists.consume.net, MOBILEMESH-LIST@lists.mitre.org
> Subject: RE: New Not "mobile mesh" released under GPL
>
> Hi to everyone out there,
>
> AFAIK, the reference to "Mobile Mesh" is for its concept not its name.
> To save confusion, some alternatives? Maybe the 'Anderson Mesh'. IMO,
a
> discussion about name seems futile when proof of application is still
> being researched. Like arguing over the name of a car before inventing
> the engine!
>
> Heck, the code has no MobileMesh content at all! It's small and fast.
> Here at CommunityWireless.org, we had a dream, and it was simple: Get
> Linux on as small a unit as possible, then tweak it to run a meshed
> network. Thanks to OpenAP (http://opensource.instant802.com) and a USR
> 2450 Router, we were half way there. We had an AP running Linux, fully
> embedded, and for less than 200UKP! Then the problem: All known
routing
> implementations are WAY too big to fit into the tiny space left on
these
> fantastic units. That included MobileMesh.
>
> So we did what any sensible group does when they hit a problem; we
> turned to a group of experts: Locust Technologies. This isn't an
advert,
> but they took a good look around, including at MobileMesh, and after
it
> became apparent that a 'description' wasn't enough, they hacked it up.
> So hj-mm was born.
>
> Now it works as-is; I have 2 USR AP's running it, and it's quick. Now
> Locust have what they believe is a working model, and so they in turn
> have turned to a larger group of experts. If everyone worked on this
> from a positive angle, and if it does scale as expected, this model is
> absolutely capable of creating the framework for a UK wide,
> multi-layered network. No it's not IP v6 (it doesn't need to be; IPv6
is
> designed for mobile networks, and we are talking about a UK wide
static
> network), but it WILL scale much further than Mobile Mesh can. It's
also
> classless! So IPv4
>
> As I understand how it works, it's dead simple really:
>
> Each node just knows which direction to send data in. So imagine a
grid:
> if you're going to a point 'north', your 'north' route will be lower
> than your 'south' route so you send that way. The next node the same.
> Its like a homing missile.
>
> In theory, circular routes shouldn't happen as the routing is dynamic.
> Scaling issues are based on the number of LOCAL neighbours not the
size
> of the network, because we assume to route it all from the begin.
>
> The only downside is propagation time, but most of the network wont
> move,
> and those that do can inform as it goes. So a node a long way away
> wouldn't know the most recent route, but it would send to last KNOWN
> route,
> and that would know the new route
>
> Does that make it any clearer? It's similar to (although not exactly
> like) the Self organising network maps in Neural nets. Simple, but
very
> effective.
>
> Sorry for the long post, but I really think that comments and
criticisms
> should be positive, and constructive. Heck, lets call it 'The Blob'
does
> it really matter?
>
> Geoff
> CommunityWireless.org
>
> -----Original Message-----
> From: R.D. Hammond [mailto:muaddib@databit7.com]
> Sent: 23 April 2002 18:32
> To: Jon Anderson
> Cc: consume-routing@lists.consume.net;
MOBILEMESH-LIST@lists.mitre.org;
> geoff@communitywireless.org
> Subject: New Not "mobile mesh" released under GPL
>
>
> Is this code mobile mesh? id you enter it into CVS with the owner?
>
> please dont steal the name. its not just about Intelectual Property
> theft. It
> causes serious headache for people when asked to insall "Claris Word"
> and there
> are 4 different pieces of software called 'claris word'!
>
> Please fix the name, thanks
>
>
> On Tue, 23 Apr 2002, Jon Anderson wrote:
>
> > Date: Tue, 23 Apr 2002 12:07:36 +0100
> > From: Jon Anderson <jon@LOCUST.CO.UK>
> > To: consume-routing@lists.consume.net,
MOBILEMESH-LIST@lists.mitre.org
> > Cc: geoff@communitywireless.org
> > Subject: New mobile mesh released under GPL
> >
> > Yesterday, I released a binary for my experimental mobile mesh
system.
> > Lots of people asked for the source...
> >
> > The biggest thing I need with this project is assistance.
> >
> > I'm worried about releasing the source with the protocol only half
> > defined, but it appears that people are unwilling to help projects
> > outside of the GPL. I can understand this, as no-one knows that my
> > intentions are really pure.
> >
> > The code is ALPHA, it demonstrates my principles and mostly
implements
> > the encoding method I described. It wont scale to production yet and
> > there are some critical parts missing, such as intelligent route
> > gathering, interface with the kernel, interproccess communication,
> > killing of dead routes, testing of peers etc.
> >
> > I'm hoping that with the help of the opensource community this can
be
> > got into a stable and high quality state.
> >
> > The code is available here:
> >
> > http://theophany.co.uk/hj-mm-01-src-gpl.tgz
> >
> > I welcome any and all comments.
> >
> > Best Regards,
> >
> > Jon Anderson
> > http://Locust.Net
> >
>
>
>


---------------------------------------------------------------------
To unsubscribe, e-mail: consume-routing-unsubscribe@lists.consume.net
For additional commands, e-mail: consume-routing-help@lists.consume.net