Noise and EDFA

[Abdallah Ismail]

If I disabled ‘convert noise bins’ and tried to get noise as a separate vector in matlab component at the output of the EDFA.
I can’t do that.
The noise was given as power value with phase equal to zero and It was different than the no. of the signal samples.
How can I get EDFA Noise Samples to use it with matlab component?Its vector length should be equal to the signal samples vector length.

[Damian Marek]

When you convert Noise bins the representation of noise is converted into a sampled signal and added to the ‘Signal’. After this operation it impossible to extract the noise.

Try this:

Place a fork component after your EDFA and on both forks use a “Signal Processing Library/Tools/Optical/Signal Type Selector” component. You can set one to Sampled signals and one to Noise bins. After the selector you can convert the Noise bins signal to a sampled signal with the Convert to Sampled Signals component and that should create to signals of equal length that represent the signal and the noise.

Cheers!

[Abdallah Ismail]

Dear Damin,
Thanks for your response but it seems like quite complex for my design.
I need to model ASE noise in the matlab component because “the mode selector” doesn’t accept noise bins in its input.
I plan to use the matlab command x=wgn(2,Sampled_signal_Vector_Length,Noise_power_in_dBm,’dBm’,’complex’)
Is this command enough and which noise power I should use,”Noise power” or “noise power at 0.1nm”?Please help.
Regards,

[Damian Marek]

That should work! There are some changes to make though and I’ll explain why. You might want to look at the attached document where there is a section on Noise Bins.

The first small problem is that the white gaussian noise generated in Matlab function will be an equal power over all frequencies, which is fine if that is what you want to model, but if you are looking at the ASE of the amplifier you will need to be able to represent different powers at difference frequencies. For each noise bin you would need to run this function to generate that frequency band’s time signal and then add them all up at the end.

For an individual noise bin I would call:

x = wgn(1,Sampled_signal_Vector_Length, Noise_power_in_dBm*bandwidth_noiseBin,’dBm’,’complex’)

Since in the noise bin signals the power is interpreted as a power spectral density, which is W/Hz. If this is too complicated to understand I don’t blame you, if you want I can spend some time on a example matlab program.

Regards

[Abdallah Ismail]

Thanks Damian,I really appreciate your support and I hope if you can help me in an example for that matlab program as you told.
Still noise bins are very confusing for me.
what is bandwidth_noiseBin?is it 13 THz or 125GHz?

Hope Your support.
Thanks and best Regards,
Abdallah

[Damian Marek]

I took some screenshots of the data from Matlab attached. Each noise bin represents a range of frequencies which is bounded by the upper and lower frequency. These values are found in the Matlab vector UpperFrequency and LowerFrequency. This range of frequencies has an associated power found in the Power vector. Two rows for the two different polarizations x and y. Finally you can also get the phase offset of each frequency, in my example they were all zero.

To find the bandwidth, you just need to find the difference in the upper and lower frequency for that noise bin. They do not always have the same bandwidth!

[Abdallah Ismail]

Thanks Damian,But the phase of the noise always appear zero not only in your case do you faced non-zero for noise before?,Also the difference between upper and lower band differ slightly.
I need to know how to measure the bandwidth in case if I’ve PM-QPSK with symbol rate 112GHz and optical filter BW. 50GHz
Regards,

[Damian Marek]

The upper and lower frequencies should differ, as the difference between the two frequencies gives the bandwidth of a single noise bin. Perhaps I’m on the wrong track here, what bandwidth are you interested in? The bandwidth of all the noise generated in your project?

[Author]

Posts