[Freedombox-discuss] This white space spectrum chip might be uninteresting

Fri Feb 15 01:59:38 UTC 2013

Hello John, All,

We have been thinking about related issues in the Serval Project.
Being based in .AU, we don't (yet) have the luxury of white spaces.
Also, as we are also targeting disaster use cases we have to be able
to work in situations where geodatabases may not be available or
accessible for whatever reason.

So our initial efforts that are being sponsored by the NLnet
Foundation are being focussed on the ISM band at 915MHz, that is
available in a number of countries (.US and .AU are of particular
interest).  This band has fairly generous rules, and is capable of
fairly long distance communications.

Our very early tests, using what we know is sub-optimal forward error
correction, and with purposely poorly placed antennae (on a couch or a
desk inside a house, similar to the kinds of places a freedombox would
be expected to be plugged in) already show that it is possible to
maintain links through O(10) houses.

This is important, because WiFi does only O(1) house, PI*r^2 = O(1)
houses covered (including your own), and so you need Almost Everyone
to join to form a mesh.

But assuming a range of O(10) houses, PI*r^2 = PI*10^10 = O(100)
houses, and so you need only about 1% up-take to enable metropolitan
networks to be feasible.  That seems much more likely than the ~100%
up-take needed to make a WiFi mesh.

The trade-off is the relatively low bandwidths that are feasible.  We
are using 250kbit, and that is about as high as is feasible.  However,
it isn't as bad as it sounds, because we can use some clever frequency
hopping and related tricks to create a "multi-threaded" radio
environment, where there can be 50 senders operating in the band for
250k x 50 = 12.5mbit total throughput, with the nice property that no
one node can consume all the bandwidth. I can provide some further
explanation on this if it is of interest.

We are using the RFD900 radios that run open-source firmware, and our
GPL Serval Mesh software on top of that to give us a fully open stack,
complete with our baked in crypto layer that the Serval Mesh is based
on, and could be integrated into the FB stack if desired at some
point.

We are beginning to make a few Mesh Helper prototypes, that combine an
RFD900 radio + OpenWRT 802.11n router and battery, and will be testing
these in New Zealand over the next couple of weeks, and can feedback
here if people would like to hear more about how it goes.  Or
alternatively, I will blog about it.  There are already a couple of
blog posts that show some of what we have been up to already, e.g.:

http://servalpaul.blogspot.com.au/2013/02/breaking-wifi-barrier-serval-mesh.html
http://servalpaul.blogspot.com.au/2013/02/building-serval-mesh-helper-device.html

If anyone is interested in knowing more, or helping us out (there is a
fair bit of radio firmware rewriting to be done to realise the full
potential), just drop me a line.

Paul.

On Fri, Feb 15, 2013 at 9:38 AM, John Gilmore <gnu at toad.com> wrote:
>> http://www.v3.co.uk/v3-uk/news/2243662/white-space-spectrum-chip-approved-for-m2m-use
>> http://www.theregister.co.uk/2013/02/13/weightless_neul/
>>
>> If this could somehow be incorporated in the Novena freedom box that Bdale is trying to whip up...
>
> (0) White Space spectrum doesn't have to be used with the Weightless
> chips or protocol.  The idea is that anybody can use it with any
> modulation that stays within the bandplan.  Whitespace does not equal
> Weightless, though they of course neglect to tell you that.
>
> (1) this is a proprietary standard, all tied up in patents.  You
> aren't allowed to know how it actually works, and you can't use it
> without paying one or many patent holders (some of whom may only
> surface years from now and start demanding royalties).  It's a classic
> proprietary radio hardware ploy in that regard.
>
> (2) this is low speed (14 megabits max -- but that's for two devices
> sitting right next to each other in a faraday cage, I'm sure) yet it
> uses large amounts of bandwidth, hundreds of megahertz).  It's
> bleeding edge, and probably full of bugs and interoperability
> problems.  It is the very first entrant in using "white space" and of
> course later entrants are likely to be more efficient, more standard,
> better debugged, cheaper, better integrated, and more open.
>
> (3) The FCC's white space stuff is a catch-22.  You can't use this
> radio spectrum until you have found out, via some out-of-band
> mechanism, both exactly where you are on earth (like with a GPS chip) and
> accessed some global database of what frequencies are available at
> your GPS locations.  This has multiple painful results:
>
>   *  You need a GPS.
>   *  You need Internet access independent of this chip.
>   *  You need to contact a centralized service, and tell it your GPS location.
>   *  Your communication can easily be censored by either pressuring or
>      spoofing that centralized service, or by cutting off your
>      Internet access so you can't find out what frequencies to use.
>      Not to mention spoofing your GPS.
>   *  Your use of the spectrum can be remotely monitored by monitoring
>      the centralized service (who checks in from what locations).  And
>      since the centralized service is proprietary and/or government
>      provided, it will probably also require some kind of "credential"
>      that proves you have paid your "licensing fees" or some such,
>      which can also be used to identify your device uniquely.  Even if
>      they *don't* feed you bad results based on your identity, they
>      will know where you are and that you are using device #
>      123-456-789-1011 there, consistently over a period of years.  If
>      you bought device 123-456-789-1011 with a credit card or had it
>      delivered to an address, those records will be available to
>      inquisitive totalitarian governments, tying the device to
>      an individual person and their whole dossier.
>
> All of these suck for a FreedomBox.  What we would want is a
> communications medium that anyone can use, anyone can build into
> devices, with public specs, that uses scarce bandwith efficiently, and
> that you can just turn on and start using immediately to talk between
> two nodes (or among two hundred nodes) that wish to communicate.
> Weightless ain't that.  WiFi is much closer, particularly as its
> early patents expire.
>
> Oh, and:
>
>   *  White space communication is illegal everywhere except the United
>      States.  While I think we certainly need protection from the US
>      government, there are also lots of other governments that are
>      egregious at suppressing basic human rights.  We should build stuff
>      that works worldwide.
>
> For more details, see the documents here:
>
>   http://www.weightless.org/media/resources
>
>         John
>
> _______________________________________________
> Freedombox-discuss mailing list
> Freedombox-discuss at lists.alioth.debian.org
> http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/freedombox-discuss