Sampling Clock Problem in OFDM

I have made a simulation of ofdm on baseband and it works fine. to
simulate the sampling clock synchronization problem, i have resampled
the baseband waveform using resample function in MATLAB.
yt = resample(yt, 1601, 1600);

Now, even with such a small offset, my receiver does not work (yt
contains 1600 samples comprising 20 ofdm symbols each of 64 point + 16
cyclic prefix = 80 samples). My questions is
i). how much performance degradation (SNR degradation) i should expect
with a given offset?
ii). i cannot perceive that what possibly can hurt ofdm when there is
a very small clock synchronization problem. If i take a time domain
waveform and takes it fft and then i just append a random sample to
the end of it and again takes its fft, the result is not different a
lot.

thanks

On Jan 22, 1:16�am, aitezaz....@gmail.com wrote:
> I have made a simulation of ofdm on baseband and it works fine. to
> simulate the sampling clock synchronization problem, i have resampled
> the baseband waveform using resample function in MATLAB.
> yt = resample(yt, 1601, 1600);
>
> Now, even with such a small offset, my receiver does not work (yt
> contains 1600 samples comprising 20 ofdm symbols each of 64 point + 16
> cyclic prefix = 80 samples). My questions is
> i). how much performance degradation (SNR degradation) i should expect
> with a given offset?
> ii). i cannot perceive that what possibly can hurt ofdm when there is
> a very small clock synchronization problem. If i take a time domain
> waveform and takes it fft and then i just append a random sample to
> the end of it and again takes its fft, the result is not different a
> lot.
>
> thanks

The offset accumulates as you move through the vector. The performance
will get progressively worse if you don't track it out.

John