cindy
Full Member
Posts: 17
|
Post by cindy on Dec 11, 2005 16:06:36 GMT -5
Does anybody ever calculate the asymptotic BER when the # of paths approches infinite for an OFDM system? Is there a lower bound for this value?
Cindy
|
|
|
Post by charan langton on Dec 11, 2005 18:00:59 GMT -5
The IFFT has no effect on the signal SNR.
OFDM does not do as well under AWGN if there is a non-linearity present in the system. OFDM has Peak to Average ratio problem which is not a friendly environment for non-linear amplification. QPSK will out perform OFDM under AWGN in this case.
Charan Langton
|
|
|
Post by charan langton on Dec 11, 2005 18:04:51 GMT -5
OK, what type of environment are we talking about? AWGN or dading.
OFDM in a fading channel will do better than QPSK in a fading channel. So in a sense you can use that a lower bound.
Charan Langton
|
|
cindy
Full Member
Posts: 17
|
Post by cindy on Dec 11, 2005 20:29:20 GMT -5
>The IFFT has no effect on the signal SNR.
I know there are several definitions realted to IFFT. The difference is the constant in front of the summation. The way the matlab implemented scales down the input signal's amplitude --- This is what my simulation tells me. Therefore, in order to keep the SNR constant after IFFT, the scaling of either the signal or the noise is needed.
Another result is that the QPSK with OFDM modulation and the QPSK without OFDM modulation has the same BER. So if I ignore the PAR problem, these two systems have the same performance in terms of their BER
Of course, the above statement only holds while the ambient noise is pure AWGN. I got the similar results via simulations.
>when the ambient noise is frequency selctive rayleigh fading: My simulation told me that, while the # of paths is about 7 or 8, the QPSK with OFDM system can achieve the same BER as it is in the AWGN. While I kept increasing the # of path, the BER goes down further. I even tried set the path number to 500, then the BER becomes very very low. So I am thinking what is the lower bound for this BER in such a system, if I assume the # of paths can approach infinite --- obviously, it will be much lower than what we could get in a pure AWGN channel.
|
|