Forums

a mini question about the "sample sheme" of OFDM

Started by daor May 25, 2006
 Hi all, I have a mini question and need yr help pls.
It is simply about the understanding of sample scheme of OFDM.
For convenience,  the parameters of a OFDM system are presented as
follows.

    .the sample period: T
    .FFT/IFFT point: N
    .the OFDM symbol period: N=B7T


  Then we obtain that,
    .the carrier spacing: 1/(N=B7T)
    .the frequency of the first "baseband" carrier: 0 Hz
    ......................second...................: 0 + 1/(N=B7T) Hz
    ......................third.......................: 0 + 2/(N=B7T) Hz

    ....                                            ....
    ......................N th........................: 0 +
(N-1)/(N=B7T) Hz


   Now we compare the sample rate, 1/T, and the frequency of the N th
carrier. It is obvious that the sample doesn't abey Nyquist Lowpass
sample theorem. So I thouth the "sample scheme" of OFDM is Bandpass
sample. Am I right?=20
   Thanks in advance!

well interesting question,
                                   will try to think out answer.
meanwhile how about, will it answer your question
[3] R. G. Vaughan, N. L. Scott, D. R. White, "The theory of bandpass
sampling," IEEE Transactions on Signal Processing, vol. 39, No.
9, September 1991, pp. 1973-1984

oh sorry this one....

M. Patel, I. Darwazeh, "A software radio OFDM bandpass sampling
receiver and the effects of aperture jitter on performance,"
IEEE Radio and Wireless Conference, RAWCON 2003, in press

This isn't quite right.  Keep in mind that frequencies that appear to
be beyond the Nyquist frequency will end up wrapping (aliasing) around
into the -ve frequencies.  So...

........................N/2........................: 0 + (N/2-1)/(N*T)
Hz
......................N/2+1......................: 0 - (N/2)/(N*T) Hz
...
........................N..........................: 0 - 1/(N*T) Hz

Also, the OFDM symbol period will be N*T + M*T, where M represents the
number of samples used for the cyclic prefix and suffix.

thanks for all you!
Phil is right. I also found my mistake these days while i was
understanding the zeros padding of OFDM. In most OFDM system having 0
high-frequency carriers to the useful carriers, the zeros are always
padded at the middle data points of date before IFFT.

Phil =E5=86=99=E9=81=93=EF=BC=9A

> This isn't quite right. Keep in mind that frequencies that appear to > be beyond the Nyquist frequency will end up wrapping (aliasing) around > into the -ve frequencies. So... > > ........................N/2........................: 0 + (N/2-1)/(N*T) > Hz > ......................N/2+1......................: 0 - (N/2)/(N*T) Hz > ... > ........................N..........................: 0 - 1/(N*T) Hz > > Also, the OFDM symbol period will be N*T + M*T, where M represents the > number of samples used for the cyclic prefix and suffix.
thanks for all of you!
Phil is right. I also found my mistake these days while i was
understanding the zeros padding of OFDM. In most OFDM system having 0
high-frequency carriers to the useful carriers, the zeros are always
padded at the middle data points of date before IFFT.

Phil wrote:
> This isn't quite right. Keep in mind that frequencies that appear to > be beyond the Nyquist frequency will end up wrapping (aliasing) around > into the -ve frequencies. So... > > ........................N/2........................: 0 + (N/2-1)/(N*T) > Hz > ......................N/2+1......................: 0 - (N/2)/(N*T) Hz > ... > ........................N..........................: 0 - 1/(N*T) Hz > > Also, the OFDM symbol period will be N*T + M*T, where M represents the > number of samples used for the cyclic prefix and suffix.
Anyway, can you explain what is sample period T? ( In some papers, they
call it "Elementary Period" ) And at the receiver, what is the sampling
rate? I was confused between T and N.T