[Freedombox-discuss] Libre Planet Followup

[Paul Gardner-Stephen]

Hi James,

On Tue, Mar 22, 2011 at 7:25 AM, James Vasile <james at hackervisions.org> wrote:
> On Tue, 22 Mar 2011 06:56:39 +1030, Paul Gardner-Stephen <paul at servalproject.org> wrote:
>> Hi James,
>>
>> > Can you tell me the largest mesh actually created with these mature
>> > bits?  I'm told there are scaling problems because the routing
>> > difficulty grows faster than the nodes (e.g. it's not O(n) but more like
>> > O(n^2).
>>
>> I am aware of ~1000 nodes, which is actually more than you need if you
>> combine mesh as the "last mile" with what infrastructure happens to be
>> available to gate traffic onto an internet (not necessarily the
>> internet) to get it from place to place on the mesh.
>> The mesh itself can be continuous and much larger, provided that each
>> node only cares about a limited number of the most local nodes.
>> This limits the routing complexity to be more or less constant for the
>> local mesh, and O(n) at worst for the wormholes between regions on the
>> mesh.
>>
>> This is exactly what ServalProject.org is implementing.
>>
>> In short, I agree with Charles, that while not yet at their zenith,
>> mesh networks are sufficiently mature to factor in and to use.
>
> That is quite helpful.  Thank you.

You're welcome.

> What if there is no gate infrastructure between mesh areas?  What are
> the limits, both theoretically and empirically?

I have reason to believe that ~10,000 nearest nodes being reachable is
possible, however I would be more comfortable by sticking to around
1,000 until we can prove the higher number, which either requires some
serious simulation, or a lot of phones/plug computers, both of which
have issues in achieving.

My back-of-envelope order-of-magnitude calculation of the data
requirement for reachability in a static mesh (but with varying link
quality over time) is approximately 2* ( (1 byte * n^2) + ( 200 bytes
* n ) ) /13 ~= (1/3)n^2+30n bytes per second.  It is very likely
possible to optimise this down further, but let's run with it for now.

This gives rough mesh-reachability (without using wormholes) limits of:

Effective Link Speed : max nodes reachable (estimated)
1mbit : 800
2mbit : 1125
10mbit : 2700
25mbit : 4200
54mbit : 6300

However I totally disclaim the actual accuracy of these figures as
they really are just general punts calculated by the seat of my pants.

They are likely to be out by up to an order of magnitude in either
direction.  Also remember that a WiFi link has probably 1/2 effective
link speed of that claimed.  So 2mbit WiFi is really more like 1mbit
data (or less).

The take home point is that ~1000 nodes is well and truly doable, and
that somewhat more may well be possible.

For comparison, where I live in a triangle of houses roughly 1/2 mile
on each side has about 400 houses, and so could have everyone with a
mesh device, and the mesh would be able to hang together by itself,
and would need only one internet connection to tie it all into the
outside world.

If you wanted to stream live voice over such a mesh, you would
probably want a few more internet connections or a few nodes placed
prominently to minimise hop-count.

Realistically, if about one in 20 homes (say, one per block) actually
had a "traditional" internet service tied into the mesh it would more
than satisfy this need, and would probably be more like what would
happen.

However, it is nice to know that in more adverse situations that the
system would still be able to work.

Paul.