[Commotion-dev] Serval Mesh Helper Device / ISM long-range meshing work

Sun Feb 24 05:00:24 UTC 2013

Hello Ben,

On Sun, Feb 24, 2013 at 6:20 AM, Ben West <ben at gowasabi.net> wrote:
> Hi Paul,
>
> Thank you for your detailed response about the UBNT M900 units, and my
> apologies for the delay in posing follow-up comments questions.
>
> I've had off-and-on curiosity about radios like the M900 products, i.e.
> mount a Nanostation M2 and Nanostation Loco M900 on the same mast, with the
> Loco powered by the M2's secondary POE port and providing backhaul for NLOS
> situations.
>
> I am assuming the 915MHz radios have better ability to handle NLOS than
> 2.4GHz and 5.8GHz radios.  That is, in a environment where you can't mount
> radios very high above the ground (10-15m max) and where there is lots of
> ground clutter (trees, buildings), would the M900 perform better?  This is
> the very benefit being touted for the nascent 700MHz TV White Space radios
> here in the US, and so I'm curious if the 915MHz band might also have some
> of that same inherent benefit over the GHz ISM bands, although to a lesser
> degree.

Talking with one of the radio guys here at KiwiEx:

1. 700MHz, 915MHz, 2.4GHz and 5.8GHz bands will all be primarily LOS.
2. Lower frequencies perform better than high frequencies, but 700MHz
is still too high to get real bending.  You need sub 50MHz or so to
get serious bending.
3. The performance difference between 700MHz & 915MHz will not be
great, for equivalent power.
4. The performance of <1GHz vs >2GHz in the face of ground clutter
will potentially be significant due to the difference in wave length.
5. Power/antennae will always make a difference.

> Nevertheless, casual snooping around forum.ubnt.com suggests that the
> maximum real throughput anyone is getting from the M900 radios in P2P
> installations (using stock firmware) is only the order of 10Mbit/s.

Yes, data rates tend to be much lower, in part due to the licensing regulations.
However, for carefully thought out networks (and this does not
necessarily mean IP networks), then even a low bit rate link can be
useful.  For example, we are pushing data over links with effective
link speeds of about 48kbits.

> Finally, addressing your comment below, I'm not sure the M900 units would
> have that much higher power consumption compared to their M2 and M5
> brethren.  Even 1Watt TX power is almost an order of magnitude less than the
> AP's overall power consumption.  Indeed, UBNT claims 6.5W max power for the
> Loco M900 compared to 5.5W for the Loco M2/M5.

Our 915MHz radios consume 4W @ 1W TX, but of course the transmit duty
cycle will be <=50%, so 2W is the realistic maximum average
consumption.  A small WiFi router like the WR703N consumes about 0.3W
by our estimates.  With power scaling, we expect average power
consumption <1W, and possibly around 0.4W -- low enough to run from a
mobile-phone sized package for a couple of days or so.

Paul.

>
> On Thu, Feb 7, 2013 at 2:22 PM, Paul Gardner-Stephen
> <paul at servalproject.org> wrote:
>>
>> Hello,
>>
>> On Fri, Feb 8, 2013 at 1:16 AM, Ben West <ben at gowasabi.net> wrote:
>> > Likewise, Ubiquiti also sells 900MHz versions of their Nanostation and
>> > Bullet access points.  I understanding these units essentially just have
>> > 2.4/5.8GHz chipsets that have been translated to a different band, and
>> > with
>> > some consequent penalty in spectral efficiency.
>> >
>> > E.g.
>> > http://www.wirelessnetworkproducts.com/locom900.aspx
>>
>> These sorts of units are quite interesting.  The range is in part due
>> to the +7.5db gain on the directional antennae at each end, and
>> perhaps more importantly, the front-to-back ratio that this affords to
>> help reject interference.  If they have done the band translation
>> "right", then the spectral efficiency could well be better rather than
>> worse, as 900MHz has substantially better propagation characteristics,
>> and over typical distances yields about +9db compared with similar
>> 2.4GHz transmissions.
>>
>> 900MHz NanoStations and similar would be a useful addition to a
>> Commotion network, and you (or we) may wish to obtain some to test
>> out, but there are a few caveats, some common to all 900MHz
>> communications, and some specific to a NanoStation type product:
>>
>> 1. The ISM 915 band is only available in the Americas, Australia, New
>> Zealand, Israel and a few other countries.
>> 2. The regulations on the use of this band, and its frequency range
>> vary substantially even among those countries. For example the US
>> requires 50 channels in the hopping scheme, while Australia requires
>> only 20, but the band in the US is almost twice as wide.
>> 3. The power consumption of the unit will be relatively high if
>> transmitting at 1W.  This is perhaps more an issue for Serval-inspired
>> use-cases where mains power is assumed to be absent.
>> 4. The higher bit-rates that the unit will use (compared to what we
>> are experimenting with) will reduce the maximum range that can be
>> achieved.
>>
>> This is why we are looking at a solution where the radio is field
>> replaceable with one that is legal wherever the unit is to be used. It
>> is also why we are looking at a hand-held portable device that you
>> just shove in your other pocket, and any phone within WiFi range can
>> make use of it.
>>
>> > I would like to assume OpenWRT and derivatives will run w/o complaint on
>> > such 900MHz units, but I'm guessing it's not that simple.
>>
>> I imagine that it shouldn't be hard to put OpenWRT on the 900MHz
>> NanoStations.
>>
>> Paul.
>>
>> > On Wed, Feb 6, 2013 at 11:00 PM, Paul Gardner-Stephen
>> > <paul at servalproject.org> wrote:
>> >>
>> >> Hi All,
>> >>
>> >> Just a heads up on some stuff we are doing at Serval that I think will
>> >> be of interest to Commotion deployments more generally.
>> >>
>> >> We are now actively working on what we call our "Mesh Helper Device",
>> >> the prototype of which will consist of a TP-LINK WR703N running
>> >> Commotion OpenWRT + servald, connected to an RFD900 ISM 915MHz band
>> >> packet radio and a rechargeable battery.
>> >>
>> >> This gives the potential for much greater range than just WiFi, as
>> >> well as offering some resistance for 2.4GHz jamming in places where
>> >> that is a problem.
>> >>
>> >> The RFD900s with current firmware in urban settings offer 5x to 10x
>> >> range compared with WiFi.  For example, I could receive a decent
>> >> signal with one radio sitting on my lounge, and the other radio with
>> >> me walking around the block, up to about 150m away -- with the signal
>> >> path obliquely through perhaps 8 or 10 houses, fences, cats, dogs,
>> >> trees and everything else.
>> >>
>> >> With good clear line of sight, and a low noise floor, ranges of 10km
>> >> or more are possible, with some evidence pointing to the actual limit
>> >> being around 80km - 120km in very ideal conditions.
>> >>
>> >> We have already managed to get Serval Mesh traffic, including some
>> >> Rhizome transfers (but not yet voice) running over one of these links:
>> >>
>> >>
>> >> servalpaul.blogspot.com.au/2013/02/breaking-wifi-barrier-serval-mesh.html
>> >>
>> >> We think a complete Mesh Helper Device, including these really nice
>> >> radios could be built and sell for <US$200.  It may also be possible
>> >> to use a cheaper compatible (but slightly lower performance) radio and
>> >> get the price down to <US$100.
>> >>
>> >> The RFD900 firmware only does point-to-point communications, but it is
>> >> entirely possible to make it do "ad-hoc" point-to-multipoint
>> >> communications.  Perhaps this is something for us to look at as a
>> >> follow on after our current work block is complete.
>> >>
>> >> We are hoping to have a couple of these with us in New Zealand in a
>> >> couple of weeks for the KiwiEx field trial. General outline of our
>> >> thinking for KiwiEx at present is at:
>> >>
>> >>
>> >>
>> >> http://developer.servalproject.org/dokuwiki/doku.php?id=content:exercises:kiwiex2013
>> >>
>> >> If there are other things people would like tested while we are there,
>> >> let me know, and we will endeavour to cover them.
>> >>
>> >> Paul.
>> >>
>> >> _______________________________________________
>> >> Commotion-dev mailing list
>> >> Commotion-dev at lists.chambana.net
>> >> https://lists.chambana.net/mailman/listinfo/commotion-dev
>> >>
>> >
>> >
>> >
>> > --
>> > Ben West
>> > http://gowasabi.net
>> > ben at gowasabi.net
>> > 314-246-9434
>
>
>
>
> --
> Ben West
> http://gowasabi.net
> ben at gowasabi.net
> 314-246-9434