Preamble detection in ofdm synchronization

> Do you mean correlating the cyclic prefix against the delayed signal
> or a specific preamble?   In either case, though, the processing is
> done at baseband so the signal is complex.

U mentioned processing is done at baseband, meaning the correlation is
done after FFT at the receiver? but most blocks i c have their
estimation and synchronization blocks before FFT. The ones after FFT
are only the estimation and equalization blocks.

What am i missing here? Should i perform the sync operation after FFT?
Shouldn't i synchronize first and get the correct start of the frame
before FFTing it? And if so, how could i perform complex conjugate on
the receiving signal which is real? I'm transmitting in BPSK...please
explain..thx...

Yes, it's sync with cyclic prefix

> Do you mean correlating the cyclic prefix against the delayed signal
> or a specific preamble?   In either case, though, the processing is
> done at baseband so the signal is complex.

U mentioned processing is done at baseband, meaning the correlation is
done after FFT at the receiver? but most blocks i c have their
estimation and synchronization blocks before FFT. The ones after FFT
are only the estimation and equalization blocks.

What am i missing here? Should i perform the sync operation after FFT?
Shouldn't i synchronize first and get the correct start of the frame
before FFTing it? And if so, how could i perform complex conjugate on
the receiving signal which is real? I'm transmitting in BPSK...please
explain..thx...

Yes, it's sync with cyclic prefix

> Do you mean correlating the cyclic prefix against the delayed signal
> or a specific preamble?   In either case, though, the processing is
> done at baseband so the signal is complex.

U mentioned processing is done at baseband, meaning the correlation is
done after FFT at the receiver? but most blocks i c have their
estimation and synchronization blocks before FFT. The ones after FFT
are only the estimation and equalization blocks.

What am i missing here? Should i perform the sync operation after FFT?
Shouldn't i synchronize first and get the correct start of the frame
before FFTing it? And if so, how could i perform complex conjugate on
the receiving signal which is real? I'm transmitting in BPSK...please
explain..thx...

Yes, it's sync with cyclic prefix

On 11 Dec 2005 16:55:01 -0800, "Ant_Magma" <vcteo1@gmail.com> wrote:

>> Do you mean correlating the cyclic prefix against the delayed signal
>> or a specific preamble?   In either case, though, the processing is
>> done at baseband so the signal is complex.
>
>U mentioned processing is done at baseband, meaning the correlation is
>done after FFT at the receiver? but most blocks i c have their
>estimation and synchronization blocks before FFT. The ones after FFT
>are only the estimation and equalization blocks.

No, I mean processing the signal after it has been mixed with a
complex (quadrature) mixer so that the spectrum straddles DC.   The
signal is no longer real-valued so that complex-valued processing can
be done without any trouble.   This eliminates the difficulty in
applying a complex-conjugate operation to a real signal.

Usually what is done then is to correlate a portion of the signal
equal to the length of the cyclic prefic with a similar section
delayed the length of the symbol.   Since the cyclic prefix should be
equal to the end of the symbol (except for ISI), a strong correlation
there will indicate a symbol boundary.

>What am i missing here? Should i perform the sync operation after FFT?

There are algorithms that work after the FFT, but usually in
conjunction with the cyclic prefix correlation I just described.

>Shouldn't i synchronize first and get the correct start of the frame
>before FFTing it? And if so, how could i perform complex conjugate on
>the receiving signal which is real? I'm transmitting in BPSK...please
>explain..thx...

I think I've covered that, but to summarize:  If the signal is mixed
to baseband it is restored to its complex-valued form, so
complex-valued arithmetic (including conjugation) works without
problems.  The symbol synchronization is often obtained using the
cyclic prefix correlation against the end of the symbol, and that is
done in the time domain before the FFT (and can therefore be used to
determine the FFT window).

This sort of thing is covered in most texts about OFDM.   If you don't
already have such a text it sounds like it might be a good idea to try
to find one.
Eric Jacobsen
Minister of Algorithms, Intel Corp.
My opinions may not be Intel's opinions.
http://www.ericjacobsen.org

> No, I mean processing the signal after it has been mixed with a
> complex (quadrature) mixer so that the spectrum straddles DC.   The
> signal is no longer real-valued so that complex-valued processing can
> be done without any trouble.   This eliminates the difficulty in
> applying a complex-conjugate operation to a real signal.

Thx for replying Eric...

This complex mixer, is it sort of like a real to complex converter? If
so how do i simulate this in Simulink? Would a "real to complex block"
work? Later on i have to convert this model to a TI C6713. If the
signal becomes complex after the synchronization, do i have to convert
it back to real before entering FFT (since i would want to get back
complex output at the FFT for the BPSK/DBPSK demodulation)?

> Usually what is done then is to correlate a portion of the signal
> equal to the length of the cyclic prefic with a similar section
> delayed the length of the symbol.   Since the cyclic prefix should be
> equal to the end of the symbol (except for ISI), a strong correlation
> there will indicate a symbol boundary.

Yes, this part is fairly simple to understand.

Unfortunately, i've read in certain papers and books such as OFDM for
wireless communications by Van Nee that synchronization with cyclic
prefix is only suitable for subcarriers more than 100. I'm only using
64 FFT with 48 data sub-c thus it wouldn't be suitable.

> I think I've covered that, but to summarize:  If the signal is mixed
> to baseband it is restored to its complex-valued form, so
> complex-valued arithmetic (including conjugation) works without
> problems.  The symbol synchronization is often obtained using the
> cyclic prefix correlation against the end of the symbol, and that is
> done in the time domain before the FFT (and can therefore be used to
> determine the FFT window).

So you're saying that the downconversion from the carrier frequency to
the baseband process itself would restore its complex value right? Then
how do i simulate this in Simulink, like i mentioned before use a real
to complex conversion block?

> This sort of thing is covered in most texts about OFDM.   If you don't
> already have such a text it sounds like it might be a good idea to try
> to find one.

I've read a few OFDM books regarding synchronization however they are
mostly theoritical with the lack of implementation explanation.

I hope you dont mind the questions, i'm just having a hard time
understanding the concept and implementing it...thx again...

On 12 Dec 2005 16:44:06 -0800, "Ant_Magma" <vcteo1@gmail.com> wrote:

>This complex mixer, is it sort of like a real to complex converter? If
>so how do i simulate this in Simulink? Would a "real to complex block"
>work? Later on i have to convert this model to a TI C6713.

This is pretty fundamental in communications, and most communications
texts should have something about baseband conversion.   Mixing with a
complex sinusoid produces a complex output at the mix frequency. It's
as simple as that, but I'll let you investigate further on your own.
As I mentioned, this is pretty fundamental so some simple library or
web searches should get you what you need here.

> If the
>signal becomes complex after the synchronization, do i have to convert
>it back to real before entering FFT (since i would want to get back
>complex output at the FFT for the BPSK/DBPSK demodulation)?

FFTs process complex values, why convert back to real?

>Unfortunately, i've read in certain papers and books such as OFDM for
>wireless communications by Van Nee that synchronization with cyclic
>prefix is only suitable for subcarriers more than 100. I'm only using
>64 FFT with 48 data sub-c thus it wouldn't be suitable.

I'm not sure why makes such a restriction.   The CP in 802.11a/g is
long compared to the symbol length so many 802.11a/g implementations
use this technique.  I've not worked the details personally.

Eric Jacobsen
Minister of Algorithms, Intel Corp.
My opinions may not be Intel's opinions.
http://www.ericjacobsen.org

Thx eric, i guess i have do more reading...