0
$\begingroup$

Is there any open and free data-set for digital ISI channels? I want to have a variety of ISI channels' taps, such as $h[k]$ here. Especially looking for BPSK channels.

$\endgroup$
  • 1
    $\begingroup$ Note that a channel model will work for any modulation -- how the signal was modulated (BPSK or otherwise) is usually irrelevant. $\endgroup$ – MBaz Sep 25 '18 at 15:42
  • $\begingroup$ @MBaz That's not 100% correct, if I use BPSK modulation I assume I don't have 'Q' channel, only 'I' so the taps should be non-complex. $\endgroup$ – Codevan Sep 25 '18 at 15:46
  • 1
    $\begingroup$ Yeah, but the point is, in a wireless channel some of the energy in the 'I' channel will spill to the 'Q' channel, so you really need both in the receiver. The taps should be complex even if you don't transmit in quadrature. $\endgroup$ – MBaz Sep 25 '18 at 16:25
  • 1
    $\begingroup$ Exactly as MBaz has stated, consider in all typical applications for BPSK there will be offsets between the transmitter and receiver carriers (hence the need for carrier recovery) so while in the channel the transmit signal will be rotating (through I and Q equally) relative to the receiver at the carrier offset. $\endgroup$ – Dan Boschen Sep 25 '18 at 16:49
1
$\begingroup$

Things like the good ole' GSM standard specify power delay profiles and a phase variation model. So, there you go, reference model delivered in the shape of a standard.

Other than that, if you're just looking for any ISI-inducing (i.e. frequency-selective) channel, why not use your favourite DSP framework to come up with a random FIR filter?

Also, a lot of frameworks do come with their own channel simulations; GNU Radio for example brings multiple channel models of varying realism in its gr-channel module.

I often find that students overestimate the complexity of actually measuring a channel impulse response to use it in their own channel simulations. Just send a PN, receive it at the same time, and use a correlation to find the impulse response; if your receiver and transmitter exhibit a frequency offset, you might need to try a couple of frequency shifts and use the one that gives the highest correlation coefficients.

$\endgroup$

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.