ODFM Question

Idiots question on basic OFDM:


Suppose I have data 1,1,1,1,0,0,0,0,1,1,1,1 and I want to convert this
to OFDM.

First of all - should logic zero be -1 instead of zero?

Also my understanding is that I convert the serial data to parallel and
then take an Inverse FFT.

Does this not generate complex data? If so do you take the real part
only or the modulus?
Also, should the parallel data be othogonal before doing the IFFT?

Many thanks

Naebad

naebad wrote:

> Idiots question on basic OFDM:
> 
> 
> Suppose I have data 1,1,1,1,0,0,0,0,1,1,1,1 and I want to convert this
> to OFDM.
> 
> First of all - should logic zero be -1 instead of zero?
> 
> Also my understanding is that I convert the serial data to parallel and
> then take an Inverse FFT.
> 
> Does this not generate complex data? If so do you take the real part
> only or the modulus?
> Also, should the parallel data be othogonal before doing the IFFT?
> 
> Many thanks
> 
> Naebad
> 
I know only slightly more than the basics, so if someone corrects me 
they're probably right.

OFDM essentially generates a phase-shift keyed signal in each 'bin' of 
your FFT.  If you want binary phase-shift keying then yes, you use 1 and 
-1.  If you want something fancier then you can use complex numbers, so 
for quadrature psk you'd use 1, i, -1, -i, for 8PSK you'd use 1, 
0.707(1+i), etc.

If you fill all of your bins with data then yes, it generates complex 
data.  This is not a bad thing if you're feeding the resulting data to 
an I/Q modulator.  If you're _not_ then you would have to be responsible 
for frequency shifting the data and keeping it real.  You could either 
do this in FFT-space by offsetting your frequency and appropriately 
mirroring it so the output of your FFT was real, or you could do your 
frequency shifting numerically after the FFT, essentially implementing 
an I/Q modulator in software before applying a single signal to a DAC.

The parallel data does not need to be orthogonal before doing the IFFT 
-- orthogonality is guaranteed by the IFFT process.

-- 

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

"Tim Wescott" <tim@seemywebsite.com> wrote in message
news:QKGdnYrFe-WldELenZ2dnUVZ_v2dnZ2d@web-ster.com...
> naebad wrote:
>
> > Idiots question on basic OFDM:
> >
> >
> > Suppose I have data 1,1,1,1,0,0,0,0,1,1,1,1 and I want to convert this
> > to OFDM.
> >
> > First of all - should logic zero be -1 instead of zero?
> >
> > Also my understanding is that I convert the serial data to parallel and
> > then take an Inverse FFT.
> >
> > Does this not generate complex data? If so do you take the real part
> > only or the modulus?
> > Also, should the parallel data be othogonal before doing the IFFT?
> >
> > Many thanks
> >
> > Naebad
> >
> I know only slightly more than the basics, so if someone corrects me
> they're probably right.
>
> OFDM essentially generates a phase-shift keyed signal in each 'bin' of
> your FFT.  If you want binary phase-shift keying then yes, you use 1 and
> -1.  If you want something fancier then you can use complex numbers, so
> for quadrature psk you'd use 1, i, -1, -i, for 8PSK you'd use 1,
> 0.707(1+i), etc.
>
> If you fill all of your bins with data then yes, it generates complex
> data.  This is not a bad thing if you're feeding the resulting data to
> an I/Q modulator.  If you're _not_ then you would have to be responsible
> for frequency shifting the data and keeping it real.  You could either
> do this in FFT-space by offsetting your frequency and appropriately
> mirroring it so the output of your FFT was real, or you could do your
> frequency shifting numerically after the FFT, essentially implementing
> an I/Q modulator in software before applying a single signal to a DAC.
>
>

I thought it was already modulated after you did the IFFT? Do you mean
modulation by the main carrier ie
take I and multiply by cos(wct) and take Q and multiply by sin(wct) - then
add?


Bob

"Bob the Builder" <FU2@yahoo.co.zpr> wrote in message
news:uaUDf.105008$vH5.1035886@news.xtra.co.nz...
>
> "Tim Wescott" <tim@seemywebsite.com> wrote in message
> news:QKGdnYrFe-WldELenZ2dnUVZ_v2dnZ2d@web-ster.com...
> > naebad wrote:
> >
> > > Idiots question on basic OFDM:
> > >
> > >
> > > Suppose I have data 1,1,1,1,0,0,0,0,1,1,1,1 and I want to convert this
> > > to OFDM.
> > >
> > > First of all - should logic zero be -1 instead of zero?
> > >
> > > Also my understanding is that I convert the serial data to parallel
and
> > > then take an Inverse FFT.
> > >
> > > Does this not generate complex data? If so do you take the real part
> > > only or the modulus?
> > > Also, should the parallel data be othogonal before doing the IFFT?
> > >
> > > Many thanks
> > >
> > > Naebad
> > >
> > I know only slightly more than the basics, so if someone corrects me
> > they're probably right.
> >
> > OFDM essentially generates a phase-shift keyed signal in each 'bin' of
> > your FFT.  If you want binary phase-shift keying then yes, you use 1 and
> > -1.  If you want something fancier then you can use complex numbers, so
> > for quadrature psk you'd use 1, i, -1, -i, for 8PSK you'd use 1,
> > 0.707(1+i), etc.
> >
> > If you fill all of your bins with data then yes, it generates complex
> > data.  This is not a bad thing if you're feeding the resulting data to
> > an I/Q modulator.  If you're _not_ then you would have to be responsible
> > for frequency shifting the data and keeping it real.  You could either
> > do this in FFT-space by offsetting your frequency and appropriately
> > mirroring it so the output of your FFT was real, or you could do your
> > frequency shifting numerically after the FFT, essentially implementing
> > an I/Q modulator in software before applying a single signal to a DAC.
> >
> >
>
> I thought it was already modulated after you did the IFFT? Do you mean
> modulation by the main carrier ie
> take I and multiply by cos(wct) and take Q and multiply by sin(wct) - then
> add?

Yes - that's what Tim means with his last statement.

Cheers
Bhaskar