[Freedombox-discuss] Allied Efforts.

[michi1 at michaelblizek.twilightparadox.com]

Hi!

On 16:40 Thu 20 Oct     , Paul Gardner-Stephen wrote:
> Hello,
> 
> On Thu, Oct 20, 2011 at 4:11 PM,
> <michi1 at michaelblizek.twilightparadox.com> wrote:
> > Hi!
> >
> > On 11:29 Thu 20 Oct     , Paul Gardner-Stephen wrote:
> > ...
> >> We are also actively engaging in the IEEE 802.11 standards process to
> >> get ad-hoc fixed (ask me if you need information about the problems
> >> with using ad-hoc WiFi to form mesh networks),
> >
> > What exactly do you mean? I can imagine supporting dealing with lots of
> > devices + drivers is a pain and I remember an email somewhere whishing for
> > lower rates for better sensitivity.
> 
> Ad-hoc wifi suffers from some low-level protocol problems that waste
> available airtime (through unnecessary beacons) and cause "cell
> splitting" through an overly complex BSSID negotiation process that
> almost every firmware gets wrong, and simply isn't needed because the
> BSSID should be fixed for a mesh.

IMHO there should not be any beacons at all for mesh networks, because the
layer 3 will do neighbor discovery anyway. But isn't this something which
could be fixed in the softmac layer for many devices? I once took a quick look
at the softmac layer. I guess there is a lot which could be optimised, if
compatibility to other operating systems is not important. For my project I
had in mind to disable retransmits and let the layer 3 handle it (and give
feedback to the layer 2).

Yes, fullmac devices will not run this way. However, at least for my project I
do not know whether I would care too much about this.

> Separately, we do wish to get lower bit rates as well, but that might
> fall into a separate standard.  But we certainly do want to go down to
> rates as low as perhaps 10kbit to get a 20db gain over regular WiFi.

Do you think 10kbit is enough for voice + route discovery + forwarding? Are we
talking about meshes?

> Using the ISM915 band would get us another 9db or so,

Yes, but only if you scale the antennas. If you reduce the frequency and keep
the antenna size is kept the same, the gain will be compensated. The signal
strength will be exactly the same. I guess you will need an antenna of ~1m
length. How will the mobile phone look like?

However, at might be helpful that lower frequencies are not blocked by
obstacles so easily.

> and the 10x power output gives another 10db, for a total 39db compared to
> regular WiFi.

Yes, this will help.

...
> >> and also to standardise
> >> the use of the cellular radio present in all phones to allow ad-hoc
> >> packet radio in the ISM915 band to allow long-range mesh networking.
> >
> > I am looking forward for this.
> 
> As are we.  But don't go holding your breath.  Fixing standards and
> seeing products arrive with the fixed standards takes years.

I did not expect anything else.

	-Michi
-- 
programing a layer 3+4 network protocol for mesh networks
see http://michaelblizek.twilightparadox.com