Post by Mike StumpBut the main problem was signal due to the huge number of wi-fi WAPs
in the neighborhood. When I checked the other day I could count 67
signals, nearly all with usuable signals.
:-( I'm still hoping that we'll evolve wifi into somehting
cooperative, mesh capable, failover capable with default public
routing to the internet.
Don't forget auto configuring, dynamic reconfiguration, self-healing,
and possibly timing compensated for large changes in range.
Post by Mike StumpI wanted that sort of protocol on 5-6
channels of the old TV spectrum and you could _only_ use it if you
routed to the internet (or mesh to someone else that could).
That would be 802.22 and 802.11af
<https://en.wikipedia.org/wiki/IEEE_802.11af>
<https://en.wikipedia.org/wiki/IEEE_802.22>
Works well in rural areas with few TV stations. Works badly in metro
areas where the requirement to protect occupied TV channels and
commercial or public safety UHF users offers only a few or no
available TV channels. You can check your area at:
<http://whitespaces.spectrumbridge.com/whitespaces/home.aspx>
Incidentally, I just blundered across a page of FCC maps. Kinda
interesting:
<https://www.fcc.gov/reports-research/maps/>
At my house in Ben Lomond, I can use 2 channels. At my office in
Santa Cruz, 4 channels. Yech.
Post by Mike StumpAlso,
cooperative power boosting, so that you can go 20 (or so) miles, if no
one else was around.
Increasing power in order to obtain additional range is futile. All
it does it "pollute" large areas with your excessive RF. Instead, a
directional antenna, aimed at your destination, is more efficient and
creates fewer problems. An expansion of the 802.11n concept of beam
steering can probably do the job, although the antennas involved will
be much larger.
Post by Mike StumpThink farm in Montana.
Right. One farm, one user, one access point, one link. You don't
need mesh, steerable antennas, or much sophistication. Point to point
MIMO links from Ubiquiti will get you 300 Mbits/sec at up to 15 miles
for little cost.
Post by Mike StumpBut, once the density of
people starts to climb, most of the units would back off on power, and
the distance you'd have to go would drop and the power needed to go
there would drop as well. Urban area with 67 stations in 300 feet, no
problem, just keep backing off on power. 1 or 2 might run at 2 mW,
but the rest, the smallest amount of power needed to keep roughly
1/nth of the clients online, where n is roughly the number of nodes
that can sink traffic to the network.
Won't work very well. It violates a method that does work. Automatic
Power Control is used heavily in all cellular systems. The TX power
is set by measuring the RX SNR (signal to noise ratio) and dynamically
adjusting the TX power for some preset SNR level where the link is
known to operate efficiently. When connected to multiple access
points, the TX power level can be different. This is NOT being done
with common wi-fi and is the cause of many interference problem.
Your scheme appears to be based upon system density, where the power
level is adjusted by counting the number of potentially interfering
stations and reducing the TX power by some formula or table. That was
tried by some company that I can't seem to recall when implementing
their proprietary polling system. The object was to maintain a
constant power density over a given area. In theory, it should have
worked, but I don't think they ever got it to work reliably. The big
problem was that the lowest cost route would change almost
continuously as new stations would arrive and disappear. Their
routing protocol couldn't handle it.
Incidentally 67 access points is nothing. If I use a passive sniffer
(Kismet), which will also show client radios and devices, I can often
see hundreds of devices. Whatever scheme you propose should be able
to also handle interference from client radios. This also becomes a
problem with mesh networks, where every node is treated as an access
point.
Take a step backwards and ask yourself WHY you've proposed various
solutions to David Kaye's problem. My guess(tm) is it's because wi-fi
isn't optimized for any particular way it is used. Long distance
point to point wi-fi has very little in common with high density mesh
networks. Such universal mediocrity has done much to improve the
overall usability, where a change to improve one part of the system,
might break other parts.
In this case, the non-optimum problem is that the hotel has a really
marginal CAT5 backhaul. The right way to do the hotel would be a very
low power access point in each room connected to a wireless switch via
CAT5 or fiber. That's unlikely to happen, so we have to invent
elaborate protocol, mesh, and topology schemes to compensate. When
you want cable, they run the coax. When you want POTS, they run the
wires. When you want intercom, they run the wires. But, when you
want wireless, they screw around with all kinds of bizarre schemes to
avoid running the necessary backhaul wires.
Post by Mike StumpOne hard thing to get right, would be enough security to both allow
mostly open access, but if bad actors come along, a way of discovering
that and knocking them out of the system. Not sure we have a solution
to this problem yet, and not sure it's even possible; but I would love
to see an existance proof it is. Maybe the IoT people will advance
the state of the art in this direction, or, maybe we burn them alive
because they are a security nightmare.
To preserve what's left of my sanity, I don't want to enter into an
endless discussion on security.
--
Jeff Liebermann ***@cruzio.com
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558