2013-12-07

Community-based internet with wireless_mesh_networks

12.5: news.cyb/net/Community-based internet with wireless mesh networks:
. there is software available
that allows our wifi routers to form meshes
with the wifi routers of our neighbors
so we can internet with them
and make use of their out-of-community connection .
... meshing makes a lot of noise
and is unreliable compared to;
better to plan point-to-point networks .

6: web: vs point-to-point wireless networks:
I'm not saying mesh networks don't work ever;
What I am saying is that
unplanned wireless mesh networks
never work at scale.
. we tried building a community-owned Internet
using solar power and wireless mesh networks
(Orangemesh grew out of this work).
After a couple years I developed a
pretty good understanding
that wireless mesh networks aren't actually a
good way to build a real network.
My colleague used to run the largest
mesh network in the world (not joking!),
and eventually his group switched to a
carefully-planned, point-to-point wireless network
[using directed parabolic dishes,
instead of omni-directional antennae?]
When you're building large systems
you want as little unpredictability as possible,
and unfortunately unplanned mesh networks
just don't deliver there.

Reason 1: Management is hard and expensive.
. you haven't lived until you've hunted down
transient connectivity problems resulting from
RF weirdness in urban areas.
Then there are network-level issues,
like traffic shaping/throttling:
the wireless channel is [very] bandwidth constrained,
so you must do extensive shaping
to ensure everyone gets fair access
to your limited resources.

Reason 2: Omni-directional antennas.
. omnidirectional antennas also receive
interference from every direction,
making the mesh network less reliable.
[ omni's are fine when you don't know
where your end points will be,
such as when moving your laptop around the room
but in a connected neighborhood
the rooftop antennae are not moving,
hence you can use less energy
and get a stronger signal
by replacing your omni' with
2 or more parabolic dishes
that target particular other dishes .]

Directional antennas allow you to
focus your RF beam directly where you want it to go,
but now your node can't communicate with
as many other nodes,
eliminating a key property of the mesh network.
[ and since Directional antennas like the parabolic
are the way to get good reception,
you might as well give up on mesh designs,
and go for a planned point-to-point .]

Reason 4: Single-radio equipment doesn't work;
multi-radio equipment is very expensive
and solves the half-duplex problem,
but you still have the interference issue,
and if you use multiple channels
to get around that problem
you quickly will run out of RF spectrum,
not to mention having the new problem
of how to intelligently allocate
spectrum to each node.
This spectrum allocation task is an
NP-hard scheduling problem,
as is allocating non-interfering
time-slots for single-radio equipment.
There are also challenging practical considerations
like how you efficiently implement
a valid schedule once you compute it.
And, because you still would need roughly
the same node density as before,
a network of multi-radio devices
quickly becomes very expensive.

Reason 5: Unplanned mesh networks break routing.
. poor RF-level connectivity means the
connectivity state between nodes changes frequently,
leading to more routing overhead in the network.
point to point doesn't need parabolic .
ubiquiti networks`airFiber uses parabolics .
. it has 10mile reach and unprecedented throughput .

nytimes`Community wireless mesh networks
 the Open Technology Institute, or O.T.I.,
at the New America Foundation in Washington,
has been the nexus for the
wireless mesh networking movement
through organizing international conferences
and funding community projects.
Last month, O.T.I. released its
Commotion Construction Kit, 
which provides step-by-step instructions
on how to set up a wireless mesh network
using open source code and
off-the-shelf routers and antennas.
To be safe, they suggest members use
a virtual private network like WiTopia or VyprVPN
on top of the networks’ baseline data encryption,
which is advisable whenever using Wi-Fi at home
or in a public space.
Free Network Foundation:
. we are hard at work on two projects:
FreedomStack is a set of tools for
building free networks,
and guifi.us is a planning, provisioning,
and crowdfunding tool
that will help people use FreedomStack.
In general, we design, implement, deploy
and talk about all sorts of network tech
that can be used to build freer networks.
Daihinia™ for Multi-hop Ad-Hoc network:
. Daihinia is a tool for WiFi.
It turns a simple Ad-Hoc network
into a Multi-hop Ad-Hoc network
which offers a higher level of flexibility
than the usual Infrastructure Mode:
in Infrastructure Mode all the computers
have to be in the range of the Access Point,
while in Multi-hop Ad-Hoc networks
they have to be within one another's range,
possibly forming chains longer than one hop .
news from the community wireless org:
As the wireless mesh grows,
many of you have been clamouring for
static IP and proper remote management of your node.
CommunityWireless has been working to this end
and is pleased to announce the
Wireless Internet Assigned Numbers Authority,
a new registry for the wireless space!
Registering with Wiana gives you
a cryptographic address certificate,
which is automatically utilised by our
current hardware / software solution.
By integrating these signed addresses into their designs,
Mesh administrators can ensure
nodes are who they claim to be!
What's more, built in to the service
is the facility to remotely manage your node,
and set up your policies and restrictions,
ensuring full node security .
[6: but every other page on that site seems dead ..]