On Tue, 22 Feb 2011 09:54:53 -0800, Jim Lux
wrote:

Here we have three (3) antennas, and as we all know they are not in
isolation.

Somewhere, there's a nearby (or near enough) overlooked reflective
surface that disrupts that oh-so-absolutely-necessary symmetry.


All practical systems like this use some form of adaptive logic to fix
that. Usually, adaptive canceling is done in the receiver, because the
signal levels are lower, but in the 802.11 kind of world, with 100mW
linear transmitters, there's probably not much cost difference.

A different matter if you're running a kilowatt.

It only takes a couple of milliWatts (kiloWatts aside) to ruin your
day in competition for listening to microWatt signals. The desired
signal's transmitter antenna would have to be literally within the
near field of the active transmitter (and receiver's) antenna system.
At that point, we may as well use a land-line with hybrid bridges.

Software coming to the rescue for a hardware problem works only in
multi-million dollar projects (fly-by-wire avionics comes to mind).

73's
Richard Clark, KB7QHC

Richard Clark wrote:
On Tue, 22 Feb 2011 09:54:53 -0800, Jim Lux
wrote:

Here we have three (3) antennas, and as we all know they are not in
isolation.

Somewhere, there's a nearby (or near enough) overlooked reflective
surface that disrupts that oh-so-absolutely-necessary symmetry.

All practical systems like this use some form of adaptive logic to fix
that. Usually, adaptive canceling is done in the receiver, because the
signal levels are lower, but in the 802.11 kind of world, with 100mW
linear transmitters, there's probably not much cost difference.

A different matter if you're running a kilowatt.

It only takes a couple of milliWatts (kiloWatts aside) to ruin your
day in competition for listening to microWatt signals. The desired
signal's transmitter antenna would have to be literally within the
near field of the active transmitter (and receiver's) antenna system.
At that point, we may as well use a land-line with hybrid bridges.

Software coming to the rescue for a hardware problem works only in
multi-million dollar projects (fly-by-wire avionics comes to mind).


In these sorts of systems (the ones alluded to in the original news
story), all the signals are in one FPGA (in digital form) and the powers
are fairly low so all the RF stuff runs basically linear (DC to RF
efficiency isn't a huge deal on a 50mW transmitter next to a 10 Watt FPGA)

So it *is* a wireline hybrid bridge.. but done with numbers instead of
transformer windings.

The point I was getting at is that in these MIMO systems, there's
already multiple receive and transmit channels with substantial signal
processing going on. So it doesn't really matter much whether you do the
cancellation/null forming in the Tx or Rx side. If you can do some
clever canceling with Tx, and make it possible to use a cheaper Rx (or,
run full duplex without needing huge dynamic range/linearity in the Rx)
then that's probably a net good.

The technique proposed is very, very similar to one used to create
increased stereo separation ("headphone sound") from conventional stereo
speakers. You send a part of the Left channel signal to the right
speaker that just cancels the signal arriving at the right ear from the
left speaker. It's a very, very impressive effect.



73's
Richard Clark, KB7QHC

On Wed, 23 Feb 2011 09:29:56 -0800, Jim Lux
wrote:

In these sorts of systems (the ones alluded to in the original news
story), all the signals are in one FPGA (in digital form) and the powers
are fairly low so all the RF stuff runs basically linear (DC to RF
efficiency isn't a huge deal on a 50mW transmitter next to a 10 Watt FPGA)

This seems to be straying from what this is about: an antenna system.
You seem to imply that the 10W FPGA is a source of radiation to
confound things. How that arrived, I don't know.

The presence of uncontrolled reflecting surfaces in proximity to the
antenna system is the objection, and the introduction of unintended
out-of-phase reflection transmission signals combined with intended
receive signals present at the receiving antenna yields the classic
problem of S+N/N degradation.

So it *is* a wireline hybrid bridge.. but done with numbers instead of
transformer windings.

This does not answer the objection. Reality (conventional usage) will
bring these corrupting out-of-phase signals and the best that software
can offer is a regression of retries to obtain error corrected packets
sorted out a the cost of seriously depressed through-put rates.

This "invention" is more about being clever than useful and belongs in
a museum case next to the Babel Fish (whose utility is questionable
when you have to listen to Vogon poetry).

73's
Richard Clark, KB7QHC

Richard Clark wrote:
On Wed, 23 Feb 2011 09:29:56 -0800, Jim Lux
wrote:

In these sorts of systems (the ones alluded to in the original news
story), all the signals are in one FPGA (in digital form) and the powers
are fairly low so all the RF stuff runs basically linear (DC to RF
efficiency isn't a huge deal on a 50mW transmitter next to a 10 Watt FPGA)

This seems to be straying from what this is about: an antenna system.
You seem to imply that the 10W FPGA is a source of radiation to
confound things. How that arrived, I don't know.

No.. the original article was talking about low power 802.11/802.16
systems (which radiate less than a watt).. At that power level, the fact
that a suitably good linear amplifier is going to consume a fair amount
of power (20% efficiency would be doing well) is insignificant next to
the power consumed by the digital processing necessary to implement the
cancellation algorithm.



The presence of uncontrolled reflecting surfaces in proximity to the
antenna system is the objection, and the introduction of unintended
out-of-phase reflection transmission signals combined with intended
receive signals present at the receiving antenna yields the classic
problem of S+N/N degradation.

So it *is* a wireline hybrid bridge.. but done with numbers instead of
transformer windings.

This does not answer the objection. Reality (conventional usage) will
bring these corrupting out-of-phase signals and the best that software
can offer is a regression of retries to obtain error corrected packets
sorted out a the cost of seriously depressed through-put rates.

No.. I would expect that this would actually work fairly well. The idea
is to allow full duplex operation, rather than the current half duplex
used in, e.g., 802.11b/g. That would double the throughput (if traffic
on the network were symmetric).

If you want to run full duplex, you have to have some way to "see" the
received signal in the face of a much larger transmit signal. Since
you're transmitting, you've got a copy of the transmit signal, so it's
really a matter of figuring out what the transfer function is from
transmitter to (self)receiver. If the external environment were fixed,
then one could probably do it with a coupler with adjustable gain and
phase. However, as you point out, the environment isn't constant, so
you need a way to adaptively adjust. You could do it with a single
transmit and single receive antenna, but that puts a tough dynamic range
requirement on the system. Say, -20dB coupling from Tx to Rx antenna,
+30dBm on transmitter, -100dBm received signal.. you need 110dB
instantaneous dynamic range.. that would need a 18-20 bit converter,
which doesn't exist at a reasonable price in the 10-50 MHz sample rates
needed.

The idea of using a pair of transmitters to reduce the dynamic range
requirement on the receiver is clever. I would think you could get 40dB
suppression without too much trouble, which gets your instantaneous
dynamic range requirement down to 70dB, which is starting to be in range
(there are inexpensive fast 14 bit converters, for instance)




This "invention" is more about being clever than useful and belongs in
a museum case next to the Babel Fish (whose utility is questionable
when you have to listen to Vogon poetry).

73's
Richard Clark, KB7QHC

On Thu, 24 Feb 2011 17:06:39 -0800, Jim Lux
wrote:

No.. the original article was talking about low power 802.11/802.16
systems (which radiate less than a watt).. At that power level, the fact
that a suitably good linear amplifier is going to consume a fair amount
of power (20% efficiency would be doing well) is insignificant next to
the power consumed by the digital processing necessary to implement the
cancellation algorithm.

As I said, how that finds its way into antenna system consideration
still remains a mystery.

No.. I would expect that this would actually work fairly well. The idea
is to allow full duplex operation, rather than the current half duplex
used in, e.g., 802.11b/g. That would double the throughput (if traffic
on the network were symmetric).

Expectation is not explanation and recital of full duplex vs half
duplex is a duplication of what has been already offered.

If you want to run full duplex, you have to have some way to "see" the
received signal in the face of a much larger transmit signal. Since
you're transmitting, you've got a copy of the transmit signal, so it's
really a matter of figuring out what the transfer function is from
transmitter to (self)receiver. If the external environment were fixed,

And there is the nut of contention. Software will have to accommodate
to variation in environment and throughput suffers. This is old
stuff. The magic of computation still admits of cost galore. More
money for poor transfer doesn't add any sparkle to market
possibilities.

73's
Richard Clark, KB7QHC

Richard Clark wrote:
On Thu, 24 Feb 2011 17:06:39 -0800, Jim Lux
wrote:

No.. the original article was talking about low power 802.11/802.16
systems (which radiate less than a watt).. At that power level, the fact
that a suitably good linear amplifier is going to consume a fair amount
of power (20% efficiency would be doing well) is insignificant next to
the power consumed by the digital processing necessary to implement the
cancellation algorithm.

As I said, how that finds its way into antenna system consideration
still remains a mystery.


because the referenced patent is talking about a *system* not just an
antenna. So the *system* implementation is relevant..