QPSK Modulated OFDM Symbol after fading channel

Question

I generate QPSK modulated OFDM signal and then send it to a EPA fading channel, after the channel, I can accurately demodulate the real part of QPSK symbol from the received spectrum, but the imaginary part is distorted so much. From what I observed from transmitted time domain OFDM symbol and received time domain symbol(absolute value), I can see that fading channel doesn't distort the signal so much. Why did the channel distort the imaginary part of the channel so much? Figure 3 is the real part of the received spectrum, figure 4 is the imaginary part of the received spectrum, after EPA fading channel and Gaussian noise addition(SNR=10dB). 8 symbols (all of them are 1+j) are located at position 40 50 60 70 80 90 100 110.

In addition, I also tried that with F-OFDM, the constellation diagram of received symbol is posted below. From the constellation diagram, I think I do not correctly do synchronization of the signal after fading channel even I tried to zero padding my signal with 7 points offset at the end before the channel and remove 7 points at the beginning of the signal after the channel. I use the MATLAB function 'lteFadingChannel'. How to do synchronization after using this function?

Answer 1

Well, a channel distorts the phase of a signal; be it a single carrier or a multi-carrier system. Hence, it's typical that each of your subcarriers has a different channel impulse response.

If you designed your OFDM system sensibly, that impulse response is typically represented by a single $h_i\in \mathbb C$; that especially means that the phase is randomly rotated.

OFDM was actually introduced because it makes equalization simple, considering that you can simply equalize each subcarrier separately by having the equalizer be individual single-tap equalizers $g_i = \frac1{h_i}$.

Now, if you're not doing equalization, then, no surprise, your signal will not be phase-correct.

You don't mention what kind of synchronization you employ, but assuming it's either something autocorrelation-based or a preamble detection: Your constellation diagram does in no way indicate you've got a synchronization problem.