Phase Noise Requirement in OFDM System!

[dopradar]

Hi all,

In OFDM system, it is strick for the phase noise requirement of LO. But how to dicide or calculate the phase noise to satisfy the OFDM system C/N requirement? Thanks!

 

[rfsystem]

Integrate the phase noise over frequency from your LO from a lower frequency to the receive bandwidth. The lower frequency depend how you frequency/phase correction for each symbol is operating. If frequency correction is done based on pilots and the S/N is high you can assume lower frequency shaping from a time discrete single pole highpass.

 

[xtasa]

RFSYSTEM,

One very easy and clean expalnation starting from natural noise which is -174 dBm/SQR Hz

For 1 Mhz bandwith is -174+10 log B=-114dBm minimum bandwith for 8 MHz bandwith OFDM TV is 10 MHz it is +10 dB which will give -104 dBm. for good receiving we have to S/N=12 dB + F(dB) of receiver .it is around -104+12+1=-91dbm that mean that we need spectral denasity of noise better than 90 dBc/sqr Hz to not see oscillator influence.We have in OFDM 6k carriers that mean that this noise have to be 1kHz from carrier.

GL XTASA

 

[spweda]

The phase noise affects the performance of OFDM in a non-linear manner. You should model the phase noise model and get the BER v.s variance of noise phase.

 

[dopradar]

Hi all,

Are there some literatures on this topic?

Thanks!
Dopradar

Added after 11 minutes:

RFSYSTEM,

One very easy and clean expalnation starting from natural noise which is -174 dBm/SQR Hz

For 1 Mhz bandwith is -174+10 log B=-114dBm minimum bandwith for 8 MHz bandwith OFDM TV is 10 MHz it is +10 dB which will give -104 dBm. for good receiving we have to S/N=12 dB + F(dB) of receiver .it is around -104+12+1=-91dbm that mean that we need spectral denasity of noise better than 90 dBc/sqr Hz to not see oscillator influence.We have in OFDM 6k carriers that mean that this noise have to be 1kHz from carrier.

GL XTASA

Hi GL XTASA,

Your analysis seems to calculate sensitivity of receiver. could you explain more about the following questions?

1)

it is around -104+12+1=-91dbm that mean that we need spectral denasity of noise better than 90 dBc/sqr Hz to not see oscillator influence

How to deduce the relation of sensitivity and oscillator influence?

2)

We have in OFDM 6k carriers that mean that this noise have to be 1kHz from carrier

Why?

Thanks a lot!
Dopradar

 

[xtasa]

Dear Dopradar,

You can find a lot of literature with a lot of mathematics to explain nature of problem you interested. My explanation is going from simple fact that you have demodulate/MIX signal than we have multiplying two signals (for beginning ideal source and non ideal carrier)and results is worse than one of each other of sources. Or other way they will carry worst noise spectral density. ODFM carriers are orthogonal and main spectral power is close to the carrier that mean that you must have very low noise carrier because you can't detect any signal correct.
Naturally this is nonlinear process but this is simple approximation to quick!!

GL XTASA:?:

 

[fasaghandis]

Search EBU technical review(www.ebu.ch). You may find good papers in this regard