I am simulating the effect of FWM and ASE noise.
finally, at the receiver
OSNR= Pout/ ( Pfwm + S * BW * m)
where the S is the ase density, m is the number of amplifier.
I have 10 channels with channel spacing 0.4nm.
So in the equation above, the BW is 4nm or BW=0.1 nm. What is the optical bandwidth reference?
please help.
many thanks
Neither.
If you are concerned about the impact of ASE and OSNR on receiver sensitivity, the dominant contribution is optical noise causing beat noise at the detector falling within the electrical bandwidth of the receiver.
For signal-spontaneous beat noise, BW is twice the electrical noise bandwidth of the receiver.
For ASE-ASE beat noise, you need to take account of both the electrical and optical bandwidths. In DWDM applications you can sometimes ignore this if the DMUX filter is sufficiently narrow, and is located between the EDFA pre-amplifier and the receiver.
Shan
Receiver bandwidth is typically 0.7 or 0.75 of the symbol rate, so around 1.8 GHz. JDSU quote 2 GHz typical for one of their APD's, so BW in your calculation would be 4 GHz.
This only applies to signal-ASE beat noise, and does not include ASE-ASE beat noise, which will depend on the bandwidth of the optical DMUX filter (around 30 GHz?) and will add a few tenths of a dB to your required OSNR.
As Reuben points out, treating FWM as noise is not always valid. The standard OSNR treatment assumes a Gaussian noise distribution. It is a useful approach when the dispersion, channel separation and symbol rate are sufficiently high for at least 1 bit walk-off between spans, and there are multiple independent FWM products falling on the target channel. It works particularly well when strong forward error correction is applied, so that the raw bit error rate is around 10-3, rather than 10-9 or 10-15.
Are you still targeting G.653 (DSF) systems with fibre zero dispersion wavelength in-band? In this case there is very little walk-off between adjacent channels, so the FWM contributions from different spans will be correlated. If your raw BER is 10-9 or higher, is may be better to treat the worst case FWM as an eye closure (distortion) penalty, rather than as noise.
A related concern with low dispersion fibre is phase correlation between spans for the FWM field. At sufficiently low dispersion, the FWM field contributions from adjacent spans are almost in phase, so that the FWM power grows as the square of the number of spans propagated rather than linearly.
Cheers
http://www.jdsu.com/en-us/Optical-Communications/Products/a-z-product-list/Pages/photodiode-apd_tia-1310_1550-nm-2_5-gbps-pigtail-receiver.aspx
http://www.kyosemi.co.jp/en/communication/ingaas_apd-tia_receiver/kpdxa2gk
- Badges
- Science topic
- Similar topics
- Computer Communications (Networks)
- Simulators