[Commotion-discuss] Mesh Networks vs Spoke Hub-Spoke Networks in Developing Countries

[Seamus Tuohy]

Hey Dan (and all),

Please excuse the delay, and the hastiness of this response. I have not
had much time to sit down to give a comprehensive reply to this
questions, but this deserves a response. As, I only have a dozen or so
minutes to write this, I want to point out that it only represents my
personal views and has not been vetted by the full team... or really any
of them.

See inline:

On 10/21/2013 03:19 PM, Daniel Hastings wrote:
> Just stumbeled across this on the Open ITP Blog:
> https://openitp.org/mesh-networks/trevor-ellermann-on-mesh-networks.html
> 
> I am curious to know the argument some of you pro-mesh people out there
> might have against Trevor's argument of Mesh networks are not the most
> suitable network infrastructure for the developing world.
> 

I actually had the good fortune of having a few conversations with
Trevor after that article went live. While I will comment more below I
think it really comes down to the deployment model, and the current
state of networking education.

Trevor has worked extensively with communities to provide internet
connectivity with limited overall access to unfiltered/any internet and
to train the individuals there on how to maintain those networks. As
such, if any part of the network goes down it is easy to pinpoint the
problem. This makes it far easier to manage a network for a user who
only has the basic networking knowledge taught to them (having spoken to
Trevor I believe that it was surely a very high quality and
comprehensive set of training they received). It also means that if the
network manager is not around to, or is unwilling to provide
connectivity to other surrounding areas those areas have no choice but
to find a "Trevor" to assist them, or go without.

He is right in many ways, but I personally believe that the challenge we
must take on is to make the construction, management, and diagnosis of
mesh networks more easily understood and taken on by individuals so that
the network can easily spread where it is needed. I think that
optimizing a network for performance should also includes creating
community guidelines about usage and educating members on how to
identify the health of their part of the network and tune their node
accordingly. But, that still requires far, far, clearer, more powerful
tools and much more education material to make possible. That is what we
are doing with the Commotion project.

Really, my argument is not a disagreement with the thoughts Trevor has.
It is a completely different intervention style based upon capacity and
the power of the network-manager. This planet has yet to see capable
individuals flocking to provide connectivity to those in need. It is far
too profitable to control avenues of communication, and far to hard to
gain the expertise and time needed to build and maintain networks at
large scale (though luckily it is getting cheaper to do). There are
groups of amazing technologists, like Trevor, who are doing amazing work
providing managed networks and the education to maintain them.  But, the
disparity is to great, and the locations to many to have people like
Trevor go out to all the communities that need him.

For these, and many other reasons, we have focused on developing tools
and educational materials that allow a community to assess their needs
and create the basic connectivity that they are lacking. We feel that
when a Trevor is not available, mesh networking has the best chance for
providing an infrastructure that will be able to grow as a community
needs it to grow. (I keep saying me because I have picked up much of
this philosophy from other members of OTI and the Commotion team, but I
should point out again that this quick overview only represents my
personal views on this project.)

> He seems to think that mesh is good in places where it can be constantly
> maintained.  I would argue that in our situation since we are not a large
> network (4-6 nodes) maintenance is not as much of an issue.
> 
> However, I can certainly understand his argument .  We do have one node
> that is not the easiest to reach (in a guard tower which means a lot of
> climbing and trying not wake sleeping guards) but for remote regions with
> many different nodes in a variety of locations troubleshooting may be a
> problem.
> 

Maintenance is such a hard problem for long standing networks. Simply
looking at the node-maps of various community wireless projects with an
eye to the version field of unconnected nodes will show dozens, if not
hundreds, of forgotten, unreachable, or otherwise unmaintainable nodes.
I have heard from many that it is just not worth the time unless they
start interfering with the network. Some of these projects are mesh,
some are managed. A house changes ownership or a store changes
management and all of a sudden a roof is off limits or an ethernet cable
unplugged without any real recourse. I think that funkfeuer and guifi
might actually have people sign contracts when they offer their roof for
usage to have some legal looking paper to hopefully convince the new
owners to at least let them reclaim the hardware.

But all that is to say that maintainance is both a hardware/software
problem, as well as a education and community engagement problem. The
Commotion project has put out, and is continuing to develop a good deal
of materials on both the actual mounting of a node
https://commotionwireless.net/docs/cck/building-mounting/prep-install-rooftop-nodes
but, also in the outreach and planning stages that will allow
communities to build the human infrastructure that will support
long-term maintenance. It is a much harder lift but, by documenting
working community models for maintaining communications infrastructure
over time it creates a more resilient human network to support a
deployed network. https://commotionwireless.net/docs/cck (keep an eye
out for the next set of planning guides which will have more of this
focus in them.)

> He also mentions the issue of power.  I've always wanted to play with the
> idea of solar powering a pico station. Has anyone seen this been done
> before? We have plenty of sun in Africa and far too many unreliable
> generators.
> 

There was a person at IS4CWN who was doing power measurements on a
pico-station to make buying the right solar panel/batteries easier.  I
am almost sure I have seen it working before as well. The Southern
California Tribal Digita Village
(http://www.newamerica.net/files/New_Media_Technology_and_Internet_Use_in_Indian_Country.pdf)
runs 5 backbone nodes on mountain-tops off of solar arrays. It is a much
more solid infrastructure, but something to look to for large long-term
solar networks.

> I am really intrigued to hear opinions about this argument especially of
> those who have worked with both types of networks or certainly have more
> knowledge on the subject.
> 
> Dan


Hope that helps,

s2e

> 
> 
> 
> _______________________________________________
> Commotion-discuss mailing list
> Commotion-discuss at lists.chambana.net
> https://lists.chambana.net/mailman/listinfo/commotion-discuss
>