A CDMA signal is considered for this application.

I would like to have an information on the generated signal oscillator's quality using a phase noise measure. If the signal was a pure carrier, it would have been easy to make a spectral measurement using a Spectrum analyzer for example to estimate the phase noise and have an information on the oscillator.

Is it possible to estimate the phase noise using a spectral measurement on this modulated signal without demodulating the spread spectrum signal ?

  • $\begingroup$ what kind of CDMA? I.e. what kind of coding are you doing? (For example,CS-DSSS would lend itself relatively nicely to that; FHSS-CDMA with actually changing oscillators would make it a lot harder. Binary DSSS might simply boil down to actually doing the despreading as the easiest way…) $\endgroup$ Commented Jun 25, 2019 at 8:01
  • $\begingroup$ Thank you for your comment Mr Muller, Actually its a GPS signal, a CS-DSSS scheme would Apply. However, I am trying to avoid demodulating the signal since the sampling clock is way less accurate than the generator's oscillator. If a Spectrum analyzer can be used to have a spectral estimation of the oscillator's instability via phase noise, my results would be less impacted. $\endgroup$
    Commented Jun 25, 2019 at 13:02
  • $\begingroup$ Well, the usual method would be completely demodulating the GPS signal in software, mathematically recovering a theoretical GPS clock. You could then, through windowed cross-correlation a recreation of that with your receive signal, get something like a phase variance, and from that infer something about phase noise power. Without de-spreading, however, your SNR is so negative that you simply won't be able to say anything about the (very low) phase noise of the GPS signal compared to the (very high) additive noise. $\endgroup$ Commented Jun 25, 2019 at 14:06
  • 1
    $\begingroup$ If you have the luxury of two independent transmitters that can be set to transmit the same sequence, and two capture channels in your receiver that are using the same inferior clock then you can cross correlate the two channels to each other instead of a known clean repllica- this will reject the local noise and the resulting measured noise from your transmitters will be 3 dB worst assuming each transmitter is similar (independent and identically distributed). $\endgroup$ Commented Jun 26, 2019 at 11:53
  • 1
    $\begingroup$ Also if your signal is indeed strong enough, and the transmitter is set to repeat your PRN sequence without data modulation, then you can assess the close in phase noise (<<500 Hz) by observing the phase noise on each resulting harmonic of the modulated waveform---The resulting waveform will have a Sinc function envelope with a tone spacing of 1 KHz. The phase noise sidebands will replicate on each of these tones, but of course overalap beyond the 500 Hz boundaries and individually be much lower signal power than you would achieve if you demodulated first. $\endgroup$ Commented Jun 26, 2019 at 11:57


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.