Baseband OFDM signal to an Intermediate Frequency

Hello,


   How is a baseband OFDM symbol upconverted to a set intermediate
frequency? Basically after the IFFT, we have carriers that are centered
around zero (DC). How can this be shifted to a frequency that is more
suitable for transmission? 

   I am trying to implement this in simulink, but am having difficulties.
Can  anyone please give me some clues/examples?

Thanks,
Rohith

Sorry I forgot to add some information:

  This is what I have tried to do thus far:

Let's say data after IFFT and Cyclic Prefix is Z = X + iY. 

For Real Data, I multiply this with cos(2*pi*Fc*t) and for the imaginary
data, I multiply that with sin(2*pi*Fc*t) and finally add these up. 

  However, when I do it, my spectral occupancy increases. I do not see my
OFDM symbol now being centered around Fc, it is still centered around zero
with huge bandwidth occupancy. 

  Not sure if any of this makes sense. So, here is a link to my test
model.

http://rohithramnath.googlepages.com/home

   Am I doing anything wrong? 

Thanks,
Rohith

  

>Hello,
>
>
>   How is a baseband OFDM symbol upconverted to a set intermediate
>frequency? Basically after the IFFT, we have carriers that are centered
>around zero (DC). How can this be shifted to a frequency that is more
>suitable for transmission? 
>
>   I am trying to implement this in simulink, but am having
difficulties.
>Can  anyone please give me some clues/examples?
>
>Thanks,
>Rohith
>

On Dec 13, 3:40 pm, "rohithramnath" <rohithramn...@gmail.com> wrote:
> Sorry I forgot to add some information:
>
>   This is what I have tried to do thus far:
>
> Let's say data after IFFT and Cyclic Prefix is Z = X + iY.
>
> For Real Data, I multiply this with cos(2*pi*Fc*t) and for the imaginary
> data, I multiply that with sin(2*pi*Fc*t) and finally add these up.
>
>   However, when I do it, my spectral occupancy increases. I do not see my
> OFDM symbol now being centered around Fc, it is still centered around zero
> with huge bandwidth occupancy.
>
>   Not sure if any of this makes sense. So, here is a link to my test
> model.
>
> http://rohithramnath.googlepages.com/home
>
>    Am I doing anything wrong?
>
> Thanks,
> Rohith
>
>
>
> >Hello,
>
> >   How is a baseband OFDM symbol upconverted to a set intermediate
> >frequency? Basically after the IFFT, we have carriers that are centered
> >around zero (DC). How can this be shifted to a frequency that is more
> >suitable for transmission?
>
> >   I am trying to implement this in simulink, but am having
> difficulties.
> >Can  anyone please give me some clues/examples?
>
> >Thanks,
> >Rohith

Clue # 1:

(x + yi)(u + vi) = (xu - yv) + (xv + yu)i.

John

>> my spectral occupancy increases
I didn't look at your file, but the problem may be that you'll have to
increase the sampling rate beforehand sufficiently.

Your output signal is real-valued, while the baseband data streams is
complex. It follows that the bandwidth must double, otherwise information
gets lost. I suspect that what you observe is aliasing.

Further, your signal gets upconverted to a carrier, and part of the
spectrum is empty. Therefore it needs even more bandwidth / higher
sampling rate.

A quick fix for the simulation is to introduce oversampling by using
larger size IFFT and zero-padding.

-mn

>>> my spectral occupancy increases
>I didn't look at your file, but the problem may be that you'll have to
>increase the sampling rate beforehand sufficiently.
>
>Your output signal is real-valued, while the baseband data streams is
>complex. It follows that the bandwidth must double, otherwise
information
>gets lost. I suspect that what you observe is aliasing.
>
>Further, your signal gets upconverted to a carrier, and part of the
>spectrum is empty. Therefore it needs even more bandwidth / higher
>sampling rate.
>
>A quick fix for the simulation is to introduce oversampling by using
>larger size IFFT and zero-padding.
>
>-mn
>

Hello Mnentwig,

>Your output signal is real-valued, while the baseband data streams is
>complex.

   Not sure which output you are referring to. My output after IFFT is
complex.

>Further, your signal gets upconverted to a carrier, and part of the
>spectrum is empty. Therefore it needs even more bandwidth / higher
>sampling rate.

   I do not follow you. Could you please explain?

>A quick fix for the simulation is to introduce oversampling by using
>larger size IFFT and zero-padding.

   This isn't possible, as increased FFT size leads to increased power
consumption.

-Rohith