0
$\begingroup$

Consider we have received a QPSK signal. We than make I (In phase) and Q (quadrature component) which I understand is nothing but multiplying the received signal by the cos(2*pi*f) and -sin(2*pi*f) Where f is the carrier frequency. Is this creation of I and Q is matched filter or there is some other filtering we do which is termed as matched filtering. I will appreciate any clarification?

Edit: Is it that we use some matched filter on I and Q data. What matched filter we use depends on the transmitter characteristic?

$\endgroup$

1 Answer 1

2
$\begingroup$

Well, you're not going to send and receive those symbols directly, you know?

What you'll need to convey the information over a channel is a waveform. Your receiver will then typically use the same waveform to correlate the signal. Say you are using root raised cosine pulses to transmit, then your receiver will employ a root raised cosine filter as a pulse matched filter to maximize the SNR.

There may be other things you can matched filter against. For instance, if you have an idea what your channel looks like, you might use a channel matched filter to maximize the SNR conveyed over your channel. Or maybe you are using spread spectrum techniques to transmit your information or things like code multiplex? You might want to use a code matched filter.

So, matched filtering can be used for a lot of things, the term alone doesn't tell it...

$\endgroup$
4
  • $\begingroup$ I am not using RRC, for example I am using simple carrier with frequency f and QPSK, in such case I get I and Q and what is the Matched filter in that case. I do not understand the difference between I and Q and matched filter. Assume no channel effect. Is it that we use some matched filter on I and Q data? $\endgroup$
    – Creator
    Jul 15, 2019 at 8:44
  • $\begingroup$ So you are using rectangular pulses? I'm surprised, due to their huge spectral leakage. But yeah I guess in this case integrating against your cos/sin carrier is what you could call matched filter. If your ADCs apply oversampling you can also do some matched filtering in the digital domain, using your I/Q samples. Totally depends on your system parameters. $\endgroup$
    – Florian
    Jul 15, 2019 at 9:17
  • $\begingroup$ Hopefully I am beginning to understand it. So, If I do RRC as you suggested, than matched filter, which is again RRC would be applied on I and Q signal. Am I correct? Would you please clarify? i would love to accept your answer and upvote. $\endgroup$
    – Creator
    Jul 15, 2019 at 9:40
  • $\begingroup$ Yes, absolutely, the pulse matched filter is something you would typically apply digitally, i.e., based on the I/Q samples. If you sample the result of this filter at its maximum point, you are guaranteed to get the best SNR. $\endgroup$
    – Florian
    Jul 15, 2019 at 10:44

Your Answer

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

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