Crafted transmision line reflections as autocorrelation discrimination and amplification

jects every mismatch except the aligned one, like this

s sequence at some fixed carrier frequency (eg, 0 = 0 degrees phase offse t, 1 = 180 degrees). There are m cycles of the carrier per bit.

received waveform at delays calculated to reflect in phase each bit of the correlation sequence according to its position. So for the 7 bit barker co de sequence

Sorry... according to how it's described, this should be something more alo ng the lines of

m?, 2m?, 3m?, 4m?+(?/2), 5m?+(?/2), 6m ?, 7m?+(?/2)

ie, the "-" bit-times are delayed 180 degrees of the carrier additionally s o the add to the amplified version.

-Andy

amplitude, when it receives the modulated autocorrelation sequence?

ically unrealistic. But I am curious if my thinking is correct.

That sort of trick has been done in radar since about the invention of dirt. ;) It's also used more generally, in most kinds of digital communications, e.g. CDMA networks and spread spectrum. In your case, you'd only win SNR if you used a threshold or some other nonlinear scheme. This is because the SNR is improved N times during the pulse and is zero the other 1-1/N of the time, leaving it unchanged on average. Thus you have to know when to ignore the signal in order to realize the improvement.

Google "processing gain" for lots more.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC 
Optics, Electro-optics, Photonics, Analog Electronics 

160 North State Road #203 
Briarcliff Manor NY 10510 

hobbs at electrooptical dot net 
http://electrooptical.net

Nonlinear or time-varying, I should say. Gating the signal on during the cross-correlation peak and off the rest of the time is the usual approach, I think--it takes awhile to figure out what the time delay is, but once you have that, you can stay in lock.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC 
Optics, Electro-optics, Photonics, Analog Electronics 

160 North State Road #203 
Briarcliff Manor NY 10510 

hobbs at electrooptical dot net 
http://electrooptical.net

On Wednesday, May 18, 2016 at 8:05:29 PM UTC+10, snipped-for-privacy@googlemail.com wro te:

jects every mismatch except the aligned one, like this

s sequence at some fixed carrier frequency (eg, 0 = 0 degrees phase offse t, 1 = 180 degrees). There are m cycles of the carrier per bit.

received waveform at delays calculated to reflect in phase each bit of the correlation sequence according to its position. So for the 7 bit barker co de sequence

amplitude, when it receives the modulated autocorrelation sequence?

ically unrealistic. But I am curious if my thinking is correct.

I think the classic form of this technique is to multiply the received sequ ence by the expected sequence and sum the totals.

A finite impulse response filter is one technique for doing this - the weig ht associated with each tap is derived from the expected sequence.

Searching on "correlation" would probably get you into the literature, whic h - as Phil Hobbs says - is extensive.

--
Bill Sloman, Sydney

I suspect that Barker codes are most useful where the detector only retains amplitude information, and discards phase information. Examples: avalanche photodiodes, and conventional photodiodes unless there is coherent mixing with a reference beam within the photodiode.

If your detector retains phase information (e.g. a conventional mixer) then I presume there are many more beneficial correlation codes available.

Yes, but the transmission lines + summing is a linear process, which is matched filtering but not autocorrelation. It will be delicately sensitive to amplitude and noise.

Reminds me of the arbitrary waveform generator used in the Nova laser, NIF's predecessor. It used machined brass transmission line things, looked like little Christmas trees, that made an arbitrary (machined in) reflections of an impulse. NIF used electronics!

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics

It is correlation, but not auto-. The FIR filter is a convolution of the input signal with the filter coefficients, and reversing the sequence of coefficients calculates correlation.

--

-TV

Hi Andy,

What Phil said +

while your signal amplitude increases by n, measured white noise increases by sqrt(n), yielding an overall increase in SNR of sqrt(n). This is no great shakes, since it is exactly what you would get by taking n independent measurements of a normal signal.

A neat feature of the system is rejection of interfering signals according to how well they (don't)correlate with your code.

Besides your communication example, these kinds of signals can be useful for making precision measurements. Measurement systems can often be modeled as communications systems - just with the information source out in the environment somewhere.

In practice you can even use randomly generated codes, or expansions of irrational numbers, or any good source of entropy. This is especially easy for measurement systems, since the transmitter and receiver are usually co-located. It's kind of neat to measure something using the digits of PI.

You've re-discovered spread spectrum modulation. It's the only signal processing scheme I know invented by a movie star.

Good search terms include DSSS and CDMA.

--

Best Regards, 

ChesterW 
+++ 
Dr Chester Wildey 
Founder MRRA Inc. 
Electronic and Optoelectronic Instruments 
MRI Motion, fNIRS Brain Scanners, Counterfeit and Covert Marker Detection 
Fort Worth, Texas, USA 
www.mrrainc.com 
wildey at mrrainc dot com

, 6m?, 7m?+(?/2)

,

SNR is generally quoted in terms of electrical power, so when you coherentl y sum N measurements of the same quantity, the signal power goes up as N**2 , and the noise power goes as N.

Cheers

Phil Hobbs

Yep, exactly ;) I thought the OP would follow better if I stayed with amplitude.

On another topic - I just noticed that AOE2 is available on Kindle. That's my new favorite book format.

Chester

0

he

? 2), 6m?, 7m?+(?/2)

of

e,

e o

is,

Thanks a lot guys for all the wise responses. I wondered if I was reinvent ing the wheel. And it's kind of clunky.

However it feels attractive because using the transmission lines "tuned ref lections" might be able to work under conditions where there is almost no p ower available.

If the transmitted PSK signal was more or less continuous, taking some time to arrive at a lock "dumbly" would be OK and once it had it, because of th e correlation characteristic it could use some active logic to gate the out put for the period until the next peak is expected.

Anyway thanks a lot everyone for the advice!

-Andy

tly sum N measurements of the same quantity, the signal power goes up as N*

*2, and the noise power goes as N.

The Kindle app for Android is great but I wouldn't buy a technical ebook. I even print specs (though I use PDFs, too).