Forums

Cylic prefix and postfix

Started by Ant_Magma November 30, 2005
Let's say my transmittid frame is like this:

|| Preamble || GI1 | OFDM symbol 1 || GI2 | OFDM symbol 2 ||

Removing GI1 would remove any interference effects caused by the
preamble.

But what if the last few samples of OFDM symbol 1 is interfered by GI2?

Wouldn't a structure like this:

|| Preamble || GI1 | OFDM symbol 1 | GI1' || GI2 | OFDM symbol 2 | GI2'
||

With GI1' and GI2' postfix able to prevent such problems? GI2' would be
also useful if there is windowing before transmission right?

I've read many papers but not many of them talk about postfixs.

Anyone can explain this?

Thx

"Ant_Magma" <vcteo1@gmail.com> wrote in message 
news:1133404518.217897.272900@f14g2000cwb.googlegroups.com...
> Let's say my transmittid frame is like this: > > || Preamble || GI1 | OFDM symbol 1 || GI2 | OFDM symbol 2 || > > Removing GI1 would remove any interference effects caused by the > preamble. > > But what if the last few samples of OFDM symbol 1 is interfered by GI2? > > Wouldn't a structure like this: > > || Preamble || GI1 | OFDM symbol 1 | GI1' || GI2 | OFDM symbol 2 | GI2' > || > > With GI1' and GI2' postfix able to prevent such problems? GI2' would be > also useful if there is windowing before transmission right? > > I've read many papers but not many of them talk about postfixs. > > Anyone can explain this?
Two reasons. 1. The guard interval is used primarily to guard against multi-path interference. Generally, this interference is a delayed version of the signal. So you'd typically have the Preamble leaking into GI1, and OFDMSymbol1 leaking into GI2, but not the reverse. 2. It doesn't really matter. If you're concerned about GI2 interfering with OFDMSymbol1, then when demodulating OFDMSymbol1, you can just back up and use part of GI1 instead of the last part of OFDMSymbol1. By backing up, you've cyclically shifted the data you're demodulating, so you need to either undo the cyclic shift before doing the FFT, or remove a phase ramp after doing the FFT. -- Eric Backus R&D Design Engineer Agilent Technologies, Inc. 425-356-6010 Tel
Regarding backing up and using GI1, assuming my guard interval is
greater than the delay spread and good frame synchronization. I suppose
i need not worry about that right?

"Ant_Magma" <vcteo1@gmail.com> wrote in message 
news:1133495543.675763.236080@z14g2000cwz.googlegroups.com...
> Regarding backing up and using GI1, assuming my guard interval is > greater than the delay spread and good frame synchronization. I suppose > i need not worry about that right?
Yes, the guard interval must be greater than the delay spread, and the difference between the two limits how much you can back up. I think a typical case would be to back up by only a small amount, perhaps to avoid any windowing between the OFDM symbols, and assume that most of the delay spread really is "delay". Even if your assumption is wrong, it probably won't be wrong by much and you'll probably still get reasonable performance from the system. -- Eric Backus R&D Design Engineer Agilent Technologies, Inc.
Thx eric!