3
$\begingroup$

I have done a base-band modeling of a 5G NR OFDM (CP-OFDM/ DFT-s-OFDM) transmission system. And I have added a DC leakage as an RF impairment and transmitted the signal through a PA (Power Amplifier).

However, at the receiver side when I try to do DC leakage correction using MATLAB's dsp.DCBlocker, some of my RB allocation (narrow RB allocation that contains the DC frequency) near the DC is corrupted and I cannot recover them correctly.
Note: I suspect this condition might be easier for LTE as LTE does not have any sub-carrier at DC.

Is there any model for DC leakage correction (LO leakage correction) that is used in practical User Equipment?

Any suggestion would be very helpful. Thank you very much.

$\endgroup$

2 Answers 2

4
$\begingroup$

Just want to provide some details to complement Marcus's answer.

(1) 5G NR does not avoid DC carrier, (2) it is possible to use it to modulate data (accepting degradation), and (3) 3GPP does not specify how to handle this.

For (1), this is because LTE devices were initially designed to support all available BW and, therefore, the DC carrier positions are a priori known, whereas, in NR, devices are allocated their (logical)-BW (via notion Bandwith part) in a very flexible way within an enormous base station bandwidth (up to 400MHz). Therefore, the Downlink DC carriers are probably outside of the UE BW. In the rare cases of (2) and (3) that DC carriers fall in the UE BW, the base station can reduce the information rate (via coding rate, allocation, modulation order, etc.) to deal with the degradation due to LO leakage. The same for Uplink DC carriers.

It is interesting to note that NR does specify the shift of half subcarrier spacing in Uplink. But this feature is used only to be aligned with LTE in the case the two standards coexist in the same place (DSS). Without DSS, we live with the DC carriers.

$\endgroup$
2
  • 2
    $\begingroup$ nice, thank you for complementing! Yeah, for NR base station hardware, there's little reason to assume the LO lies in the middle – doing a single-step direct downconversion isn't that common for NR bands, anyways. In labs of people designing >= 28 GHz OFDM transceiver hardware, I've seen basically anything – from mmWave -> ca 2 GHz IF direct sampling, to quadrature mixing to I and Q both being IF signals themselves, which then get sampled using one standard quadrature downconverter each (so, yielding basically II, IQ, QI, and QQ as baseband signals). $\endgroup$ Aug 22, 2022 at 10:44
  • $\begingroup$ @AlexTP you are right that 3GPP does not specify anything regarding LO leakage correction. I understand from your explanation that LO leakage correction depends on the designer. $\endgroup$ Aug 23, 2022 at 0:33
4
$\begingroup$

In direct conversion transceivers, the LO leakage is pretty much a given. Some systems, mainly OFDM systems like Wifi, deal with that by simply leaving the DC carriers unused. And that's it. No further treatment is neeeded.

The same happens in 4G downlink, and I suspected¹ it stays the same in 5G, at least for the non-mmWave frequencies: The DC carrier remains unused, and no DC offset cancellation is necessary.

In the uplink, there's no DC problem, because by design, this whole OFDM system is shifted by half a carrier spacing, so that LO leakage ends up on a null of all subcarriers.

So, user equipment doesn't do much DC offset cancellation in the digital domain. If anything, there might be slow control loops that calibrate out possible ADC and receive chain offsets.

(now, being honest here: it's impossible to know what every user equipment there is does. There probably are user equipments that do some additional DC offset cancellation, as part of a digital predistortion / compensation scheme for a lot of the, especially nonlinear, imperfections of their RF chains. Will they be a single building block "leakage removal"? Probably not, as that, as described, is already done by OFDM principles by design.)


¹ but AlexTP points out that this is not the case for NR

$\endgroup$

Your Answer

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

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