OFDM divides the signal in frequency domain into orthogonal and relatively narrow subcarriers where fading can be considered to be flat Flat fading is not associated with delay spread and is corrected in frequency domain

Delay spread caused by multipath is associated with frequency selective fading and in OFDM it is absorbed by the cyclic prefix which isolates consecutive OFDM symbols

How would delay spread show up in the frequency response of the channel? would it show up like "nonflat" fading or a notch within the relatively small bandwidth of some subcarriers?

i.e. it would be wrong in this case to consider the affected subcarriers still experiencing flat fading?


Because of wide-band nature of OFDM signals they are so prone and sensitive to multi-path delayed channel. The problem is that at the receiver side you will receive several copies of original transmitted signal, each with a different magnitude and delay. This "delays" are so destructive at the receiver side. The mathematics behind this: imagine the channel with multi-path problem. so u can model this channel as a linear-system with an "impulse response". this impulse response consists of several impulses with different magnitude and delays. so, your sent OFDM signal passes through this channel. the result can be guessed. at the output this system, the OFDM signal is convolved with the impulse response of the channel (system) in time domain. In frequency, it is like as the freq-domain OFDM signal is multiplied by the frequency-response of the channel. That's why we face a distorted received signal. At the receiver, when you apply FFT on the received signal you have to compensate that errors. This is called channel estimation (means estimating the freq-response of the channel). Then you should correct the signal. This is so crucial because in OFDM we usually have QAM modulation & demodulation. and in QAM the magnitude of signal is important which is already distorted by the channel.

  • $\begingroup$ You are talking about the inband amplitude distortion in general caused by multipath and I am talking about the difference in amplitude distortion between the impact of flat fading/subcarrier and frequency selective fading over the whole signal bandwidth or are they essentially the same? $\endgroup$ – ali khalil Jun 15 '20 at 21:31
  • $\begingroup$ what do you mean by "flat fading over subcarrier"? The overall fading over the OFDM bandwidth will be flat only if there's only "one" propagation route. $\endgroup$ – Hamid R. Tanhaei Jun 15 '20 at 22:45
  • $\begingroup$ So if there is more than one propagation route (multipaths), I assume that fading will be flat this time not over the whole OFDM bandwidth but over the bandwidth of a subcarrier, i.e. only 15 KHz in LTE, right? $\endgroup$ – ali khalil Jun 16 '20 at 18:28
  • $\begingroup$ Yes, u r right in a sense. But even that so-called flat-fading in a subcarrier would change not only the magnitude but also the phase of its values. And when you do FFT at receiver side and want to demodulate that data through QAM (which is based on mag & phase), the result would be full of errors. So, channel estimation is a must in OFDM. That's why "pilot" subcarriers are used in OFDM modulations as a reference to extract the channel frequency response and then compensate and correct the distorted data. $\endgroup$ – Hamid R. Tanhaei Jun 17 '20 at 10:49
  • $\begingroup$ Yes of course , assuming that channel estimation is more or less accurate enough. Thank you $\endgroup$ – ali khalil Jun 17 '20 at 19:16

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.