[hpsdr] I didn't think it would happen again "PING"

[jeff millar]

Murray Lang wrote:
> What I need are the TALENT and the TIME modules. 
...don't we all...
> BTW, I realised that the phase shifting can't be done at I/Q level because 
> that's just the modulation. 
Not so fast. With most sampling schemes, anything can be I/Q. Another 
way to say it, "meet the Nyquist rate and nothing is lost, all 
possibilities remain".
> Would shifting the LO arbitrarily be any easier 
> than the outboard scheme that I hinted at? Anyone?
>   
With direct sampling, the LO doesn't exist in analog, it moves into the 
DSP. In general, SDR uses digital because it make all forms of signal 
manipulation easier. A beamforming system will definitely prefer to 
process the signals in the DSP domain.
>> -----Original Message-----
>> From: hpsdr-bounces at hpsdr.org [mailto:hpsdr-bounces at hpsdr.org] On Behalf Of
>> Murray Lang
>> Sent: Monday, August 21, 2006 9:15 PM
>> To: hpsdr at hpsdr.org
>> Subject: Re: [hpsdr] I didn't think it would happen again "PING"
>>
>> ***** High Performance Software Defined Radio Discussion List *****
>>
>> The beam forming had occurred to me as well, but I was interested in the
>> gain. With lot/block sizes shrinking and everyone becoming more precious,
>> towers with HF beams are now virtually impossible to get approved. However
>> four verticals might be more achievable. Each could be fed with a
>> controlled phase shift to give roughly 6dB gain (I would have thought).
>> Each could also have its own PA, making higher power easier to achieve in
>> some ways. For mobile operation the control unit could be fed with a gyro
>> to correct for the vehicle/vessel orientation.
>>     
The biggest benefit of beamforming will be interference nulling rather 
than gain. As you say, a few antennas can proved 6 dB of gain. But they 
can provide multiple 30 dB nulls on strong interference.

The really cool trick comes when the DSP simultaneously form beams and 
null to optimize reception of _all_ the signals. The receiver 
effectively has multiple audio outputs, each optimized for a separate 
signal.

The HF signals intercept people lay out an array of very short whips. 
Short whip antennas provide an omni-directional receiving pattern. An 
array of them laid out in two dimensions can form beams and null 
interference in both azimuth and elevation. Transmit is a harder problem.
>> I suppose this could be done with four synchronised receivers transmitters,
>> with the phase adjustment performed at I/Q level. However, would it be
>> feasible to have  outboard units based on the LT2208, a D/A of some sort,
>> an FPGA and a big memory stick? I imagine the memory speed would limit such
>> a device to HF operation.
>>     
Present 802.11 data links have enough bandwidth to link a bunch of 
remote receivers back to a central DSP. Image a small battery (solar?) 
powered box with a whip antenna and an 802.11 USB dongle. Lay out as 
many as you can afford around the neighborhood in a more or less random 
pattern. Each receiver digitizes about 10 KHz of the band and links it 
back on about 400 kbps of data rate.

The central DSP commands the receiver to tune to a selected section of 
the band and provide both time and frequency references via the 802.11 link.

The advantage of this approach becomes economy of scale. The small 
remote receivers get produced in enough quantity that costs come down. 
The 802.11 link builds on existing massive consumer volume. Most of the 
DSP remains in the single central location. The function of the remote 
receiver remains rather simple and reliable, not likely requiring much 
software or hardware revision.
>> Just a (apparently unoriginal) thought.
>>     
Yeah, that happens all the time. I went looking for information of 
automatic beam forming and interference nulling and it's all over the 
literature.
>> At 10:16 AM 19/08/2006, jeff millar wrote:
>>     
>>> Here's some ideas:  Extend the SDR concept to multiple antennas with
>>> digital beamforming.  Some snippets of ideas...
>>>
>>>     * Four channel receiver with each channel connected to a vertical in
>>>       a four square.
>>>     * Remote receivers that use 802.11 to link sections of the band back
>>>       to the DSP, such at 3780-3800 KHz
>>>     * Rx with 802.11 has whip, preselector, preamp, A/D, DSP, uproc, USB
>>>       802.11 dongle, battery powered, clocks synchronized by 802.11
>>>       messages.
>>>     * Networked receivers, like IRLP/Echolink on steroids, link 10-20
>>>       kHz of passband in I/Q and let DSP tune with the band, DSP can
>>>       also combine signals from distributed receivers
>>>     * Lay out an array of remote receivers with small whips and beamform
>>>       using all the elements of the random array.
>>>     * Combine beam forming and interference nulling to maximize SNR
>>>     * With an array of whips, null interference in both azimuth and
>>>       elevation
>>>     * For all the signals in the passband, separate them by direction of
>>>       arrival and simultaneously beamform and interference null on _all_
>>>       signals.  Then demodulate all signals.  The output of the receiver
>>>       becomes separate, selectable, audio channels for each signal.
>>>
>>> For the mathematically inclined, the paper below suggests some DSP
>>> approaches to beamforming on all signals simultaneously.
>>>
>>>     www.nps.navy.mil/asilomar/asil03/prog03/Asilomar-Keynote.pdf
>>>
>>> Look at the slides on "blind signal separation" and "principle component
>>> analysis".
>>>
>>> jeff, wa1hco
>>> ____________

 1156255456.0