Signals with the same frequency spectrums but different power levels

[PainterGuy]

Hi,

I understand that if there are two signals whose spectrums consist of different set of frequencies, they can be separated using a filter such as bandpass filter on the receiver end. Is it possible to separate two signals who frequency spectrums are the same but power levels are different?

 

[Baluncore]

Possibly. But not using frequency domain filters.

If the higher power signal was more predictable, then you could use phase information to cancel part of it's energy. Hopefully, the low power signal will be present after the partial cancellation.

There is also a possibility that many “spread-spectrum” signals can share the same frequency band and transmit the same spectrum, but be different in their time and phase. Multiple channels can then be separated and detected by knowing the different "sequence spreading codes".
Examples are mobile phones and digital TV.
https://en.wikipedia.org/wiki/Spread_spectrum
https://en.wikipedia.org/wiki/Direct-sequence_spread_spectrum

 

[anorlunda]

The OP does not specify analog or digital signaling.

As @Baluncore hinted at: carrier sense multiple access with collision detection. It is the basis of Ethernet. In effect, it becomes a kind of time division multiplexing.

If you detect frequency f1 that's one signal. Detect f2, that's a second signal. Detect f1+f2, that's a collision, discard and retransmit.

 

[sophiecentaur]

Is it possible to separate two signals who frequency spectrums are the same but power levels are different?

The most obvious example is the FM Capture Effect. An FM discriminator (demodulator) will have an output signal that's highly dominated by the stronger of two received signals. This effect is particularly strong for wide deviation signals where the high level wanted signal phasor is whipping round in phase by many cycles each way and the low level interfering signal only produces a slight phase perturbation on that phasor. That increases the ratio of programme amplitude by many dB. It's the same principle that makes the (Wideband) FM signal to noise ratio so much better than regular AM.

Probably not what you had in mind but that fits the bill. There are many other more sophisticated methods of doing a similar thing.

 

[Baluncore]

The most obvious example is the FM Capture Effect.

A 12 dB advantage dominates the smaller signals and noise.

Probably not what you had in mind but that fits the bill.

But how can the smaller signal then be extracted and detected ?

 

[tech99]

If they have the same spectrum the weaker one cannot be extracted.

 

[sophiecentaur]

If they have the same spectrum the weaker one cannot be extracted.

Depends on the detail and how similar / same the spectra are.

I was wondering if the ( captured and cleaned up) programme on the higher signal were extracted (easy) and that signal used to modulate another carrier (same deviation but inverted). That carrier with just the one mod signal could be mixed with the original signal to produce a carrier of twice the frequency of the original with - to a first order - the higher signal mod canceled and the original modulation of the lower level signal remaining.
It's bound to have been tried by someone before DSP was available. I think it's valid.

 

[sophiecentaur]

But how can the smaller signal then be extracted and detected ?

Some people are never satisfied.
See above for an idea.

 

[Joshy]

It sounds like to me you’d be interested in reading about coding schemes like CDMA, but different power levels are usually not good.

edit:

I didn't read the message in #2 as I was poking around on my phone. Balun already got it.

 

[PainterGuy]

Thank you very much, everyone!

Actually this question came to my mind while I was reading about cellular phone system where co-channel cells operate on the same frequencies but it is required that a sufficient distance is kept between co-channel cells. It kept me wondering that even when one cell is at some distance, its transmitted signal won't completely die away. There would be at least some interference. I think this is what @Baluncore hinted at.

Probably not what you had in mind but that fits the bill. There are many other more sophisticated methods of doing a similar thing.

Actually this is more like what I had in mind. I'd say it a simple example and easier for a person like me to understand.

I couldn't find any source where it explains how capture effect effect really works. I mean how FM demodulator separates the signals based on their amplitudes. I see that you tried to explain it a little but I couldn't get it. Anyway, I'll try again to find some source which tries to explain it how FM demodulator does capture effect. Thank you!

Helpful links:
https://www.its.bldrdoc.gov/fs-1037/dir-006/_0844.htm
https://www.rfvenue.com/blog/2014/12/15/making-use-of-the-capture-effect

 

[sophiecentaur]

I couldn't find any source where it explains how capture effect effect really works. I mean how FM demodulator separates the signals based on their amplitudes.

That will be because you were looking for 'easy' search terms. You need to get inside the subject to find more and to recognise what's actually being talked about. Alternatively you could google FM Capture Effect and find a Wiki article with pictures.
No pictures from me here but the way it works is explainable by looking at phasors. An FM discriminator is basically following the phasor of the received signal. FM will rotate the phasor in either way, according to the mod signal. For wide deviation FM, the phasor will go many turns in each direction, back and forth. If you add noise, that will involve extra carriers, producing a 'fuzzy ball' round the end of the phasor and the demodulated noise will come out as phase noise of just +/- a few degrees. This will compare with the wanted signal which will correspond to 360 times a lot of rotations. That gives a 'fm noise advantage' because you have spread the carrier over many times the mod signal bandwidth. This noise is wide band channel noise but the receiver filters out all but the audio band noise from the audio output. FM falls down catastrophically once the received Wideband rf noise level takes the wanted phasor back near the origin. Then the discriminator will start to follow the noise content and the demodulated noise will dominate. CRASH!
If there is low level interfering signal, the same thing will apply. The discriminator will follow the phasor of the wanted signal and minimise the effect of the interference. Once the interference gets greater, the discriminator will follow that. With AM (or narrow band FM) the demodulated signal will comprise both signals in roughly the same proportions as the ratio of the carriers.

 

[PainterGuy]

Alternatively you could google FM Capture Effect and find a Wiki article with pictures.

Thank you!

I did check Wikipedia but it doesn't explains much: https://en.wikipedia.org/wiki/Capture_effect

I also checked some other resources, the same thing. They fall short of explaining it. They just describe it.

At least it's good to know that separating same frequency signals based on power levels could be done.

 

[sophiecentaur]

At least it's good to know that separating same frequency signals based on power levels could be done.

Certainly - under the right circs. Ordinary AM is one of the hardest examples and you'd need to have some side information about the natures of the two signals.

 

[sophiecentaur]

I did check Wikipedia but it doesn't explains much:

That is one extreme. This link is a Master's dissertation and gives it the full works. He models the wanted and unwanted signals and, afaics, he shows that the output of the demodulator (in the bandwidth of the mod signal) consists of the wanted signal (ideal operation) plus a lot of "impulse-like" spikes, due to the interfering signal. The low pass filter reduces the high frequency spikes and leaves you with a cleaned up version of the wanted modulation.
The essential idea (as with the noise improvement) is the fact that the wanted mod signal is all in the zero to 'audio' bandwidth but the unwanted or noise signals are all over the zero to full channel bandwidth and can be largely filtered out.

 

[tech99]

With wide band FM the ratio of signals to achieve capture can be very small, such as 1dB. I believe the limiter removes much of the interfering energy, which even for another FM transmission, appears as AM noise. This noise is then largely removed by low pass filtering,

 

[sophiecentaur]

I believe the limiter removes much of the interfering energy,

The limiter is not as 'clever' as that, I think, although the graphics can make it appear that way. I remember my Dad explaining it that way about the noise fuzz can be wiped off. What he didn't mention was that there's still phase noise which affects the zero crossings (which the discriminator is actually looking at). The discriminator 'follows' the relatively slow deviation of the wanted signal (the constructive result of the whole set of wanted fm sidebands) but the interfering unrelated signal will cause (as the link above says) impulses of phase which are demodulated as high frequency products which are not in the base band bandwidth - they are spread all over and the LP filtering eliminates most of that interference and the discriminator more or less 'ignores' these splats.
The interference becomes more noise like (in the same way that quantisation by an ADC produces the noise-like 'quantisation noise' - which is really distortion and not noise.)
The simulation (only a simulation) suggests that the capture effect can still work with very close carrier ratios.

 

[marcusl]

Signals in the same band are easily separated when you know the waveform you are looking for, by using a matched filter. This is the basis for CDMA, which was used successfully for cellphone communications in the US for decades and which is used for GNSS (GPS, Galileo, etc.) today. The signal you are not looking for simply increases the noise floor.

 

[sophiecentaur]

The signal you are not looking for simply increases the noise floor.

When I first heard that idea, I was chuffed to death with it. It's so elegant that each user more or less hears the other users as noise.