I just want to obtain some general ideas for such a scenario: if the environmental noise is extremely strong and the audio signal at the receiver is extremely weak(for instance, low range propagation), so even after pulse compression(cross correlation using linear chirp signal), the correlation peak is hidden in the noise. Are there any ways to detect the hidden peak?

  • 2
    $\begingroup$ With a single pulse operation, I don’t believe there is. With multi-pulse operation you can coherently add signals together to increase target response. Are you considering multi-pulse solutions? $\endgroup$ – matthewjpollard Sep 11 '18 at 22:57
  • $\begingroup$ Thx, can u plz be more specific about the multi-pulse solution? $\endgroup$ – WSL Sep 12 '18 at 2:00
  • $\begingroup$ or, can we consider this problem is equivalent to "generic finding hidden peak" problem? Cuz there should be lots of methods of doing this, including the most recent machine learning feature extraction algorithm $\endgroup$ – WSL Sep 12 '18 at 4:32

So this answer is coming from experience with radar systems which use complex signals (in-phase and quadrature signals so that phase information can be recovered), be aware that you may have to adapt it to fit your exact needs:

With radar systems, this is a pretty common problem. Sometimes, for a variety of reasons, the target that is present can’t rise above the noise, even with regular pulse compression gain. In this case, most systems turn to some sort of multi-pulse solution; a common one is called coherent integration. Basically, you send out and receive the exact same pulse N many times.

After doing this, you’ll have N many versions of the same target. Assuming no target migration has happened, you simply add all of the complex signals together. The idea here is that the target signal will cohere across all of them, such that a processing gain is achieved, but the noise will not cohere with itself and it’s power level will remain fairly constant.

  • $\begingroup$ Thank you very much for your help, I can learn from your answer that the basic idea is still to increase the SNR,is that correct? $\endgroup$ – WSL Sep 21 '18 at 14:15
  • $\begingroup$ Correct: you gain SNR through coherent integration of the multiple pulses. The phase coherency is very key here to pushing the signal above the noise, it can’t just be an amplitude integration. $\endgroup$ – matthewjpollard Sep 21 '18 at 17:37

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.