In optical OFDM systems, called DCO-OFDM, we usually use the data with its flipped complex conjugate to produce real time-domain signal while the first and $N/2 +1$ subcarriers are kept null, $N$ is the number of subcarriers in each OFDM symbol (time-domain signal is the data resulted of multiplication with inverse fast Fourier transform).

Those two subcarriers (I mean the first and $N/2+1$) are usually called DC components. What is confusing me, do these two subcarriers are forced to zeros only to have real time-domain signal, or they must be always zeros for other reasons?

For example, if we are using Hartely transforms or Cosine transform instead of fast Fourier transform to achieve OFDM system with PAM modulation, will the first and $N/2 + 1$ subcarriers must also be forced be zeros or that is not needed anymore?


1 Answer 1


The DC bin is nulled in OFDM to allow for direct conversion quadrature sampling architectures in either the transmitter or receiver, specifically sampling the complex baseband “I” and “Q” waveform and using a analog local oscillator (LO) frequency that is centered on the waveform (at the “DC bin” location in the frequency domain), for either and analog IF or analog RF. Such analog approaches inevitably have a non-zero LO feed through which produces energy in the DC bin, or adjacent bins in the receiver if the sampling is done prior to LO frequency offset correction.

If a “digital IF” approach is used for the transmitter and receiver, in which case the analog LO is outside of the signal bandwidth, then such DC nulling would not be required.

  • $\begingroup$ so in general, leaving the first and $N/2 + 1$ subcarriers to be null is not mandatory. Is that statement right ? In other words, coming back to my question, when using real transform with optical communications, we can transmit data in all subcarriers, right ? $\endgroup$
    – Sajjad
    Commented Feb 14, 2023 at 2:48
  • $\begingroup$ Yes, if you are designing your own transmitter and receiver. (And not doing any direct conversion at any stage in the down-conversion process in the analog domain-- basically no local oscillators centered on your channel). $\endgroup$ Commented Feb 14, 2023 at 2:59
  • $\begingroup$ OK, thank you .. Just to confirm, even if I work with real tests? I can normally ignore adding those zero subcarriers ? $\endgroup$
    – Sajjad
    Commented Feb 14, 2023 at 5:13
  • 1
    $\begingroup$ If the signal is real then the "positive freq" and "negative freq" bins are redundant and you can ignore all the negative bins, (or populate just the positive, and then the negative with their complex conjugate if you are taking an IFFT of it to transmit). But this does not impact whether you use the DC bin or not. Consider if you occupied the DC bin or not what you will see in the IFFT: with the DC bin occupied, you will get a DC offset. This is how the DC bin maps to your time domain waveform and that doesn't matter if that waveform is real or not. $\endgroup$ Commented Feb 14, 2023 at 12:50

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.