Forums

OFDM symbol genration - IIFT concept

Started by sandeep kumar June 8, 2005
Hi all
I have basic question regarding OFDM symbol genration.IFFT is taken
to genrate a OFDM symbol.that part is undersood.

My question is that whether that IFFT taken is complex or real.i
believe it is complex as the ouput is shown to be assymetrical.so if
it is complex,the data to be transmitted also has to be complex.

from where we fill the imaginary aray to be fed to the IFFT,while i
feel real part is filled withe the real data to be transmitted.

thanks in advance.

sandeep


Hi Sandeep,
you could modulate the signal say by M-ary QAM. This will result in
complex values & this could be fed into the IFFT block. But before
this, you will have to perform the S/P conversion.
Cheers!
Bavithi.

On 6/8/05, sandeep kumar <sandeepk1980@sand...> wrote:
> Hi all
> I have basic question regarding OFDM symbol genration.IFFT is taken
> to genrate a OFDM symbol.that part is undersood.
>
> My question is that whether that IFFT taken is complex or real.i
> believe it is complex as the ouput is shown to be assymetrical.so if
> it is complex,the data to be transmitted also has to be complex.
>
> from where we fill the imaginary aray to be fed to the IFFT,while i
> feel real part is filled withe the real data to be transmitted.
>
> thanks in advance.
>
> sandeep >
>


> from where we fill the imaginary aray to be fed to
> the IFFT,while i
> feel real part is filled withe the real data to be
> transmitted.

It's not necessary that the data to be transmitted is
real!

What's the input to the IFFT? Typically, it would be
the output of a digital modulator. If you use anything
other than BPSK as your modulation scheme, you would
end up with complex symbols, right?

In wireline DSL, they do it in a slightly different
way - the data to be transmitted is real, so the input
to the FFT is taken as the data + its conjugate. This
way the FFT output becomes real.

HTH,
K.
__________________________________________________________
Free antispam, antivirus and 1GB to save all your messages
Only in Yahoo! Mail: http://in.mail.yahoo.com



Hi All,

I have 2 major doubts:
1)In the process of generating OFDM symbols,I try to use Sinc filters
at the Tx & Rx end for pulse shaping.Now,I understand that convolution
of Sinc with a Sinc is a Sinc,however,when I upsample by 4,every 4rth
sample of the convoluted Sinc doesn't fall at exact 0 on the
X-Axis.So,as a result,the demodulated(QAM in my case) doesn't converge
at the same point & results in scatterd pattern(Even without
noise).However,if I use a single Sinc,there is no such problem.
I would like to know if it is a problem with the conv function in
Matlab or conceptually,this is hwo the conv of 2 sincs work? If yes,
what can I do to rectify this?
Also,if I use sqrt raised cosine filters on both ends,I want to work on
a smaller roll-off(say 0.009),however,the result isn't good.So,I'm
using 0.5 as of now,but I would appreaciate it if anyone can throw some
light on this one.

2)I'm trying to study ACE(Active Constellation Extension) in OFDM
systems to reduce PAPR. I've tried to go through a few
papers,however,there seems to be too much math involved & I'm just
looking for that one paper which can help me start off with this
subject...as in basic understanding of how extending the constellation
can help in reducing PAPR etc.

I appreciate any help.

Thanks!
Bavithi.