Some questions in building networks and creating arbitrary spectral information

[xiaotianj]

Hi, I want to simulate the OSNR in a real-world optical networks. When using GNpy to build networks and create spectral information, I have the following questions:

1. The build_network function requires pref_ch_db and pref_total_db. The nominal power displayed in the network management system is 1dBm, but only about 20 channels (out of 80 channels) are used, the average channel power should be about -1dBm. How should I set pref_ch_db?
2. In the examples of GNPy, all channel powers are set to the same value. However, I noticed that create_arbitrary_spectral_information function allows me to create spectral information with varying channel powers. Does this non-flat power spectrum simulate the same way as the flat one? Can tx_osnr be set to different values for different channels?

[xiaotianj]

Another question is when I set different tx_osnr, the simulation results seem to be the same. For example, when I set tx_osnr = 25, results show the rx_osnr is over 30. Actually, rx_osnr is about 22.

[EstherLerouzic]

The build_network function requires a reference channel (v2.13), but I understand the confusion: the documentation has not been properly updated yet, thanks for pointing this!. The reference channel is used to determine the total power of amplifiers (full load) and then compute the design.

If your question is about designing the network for 80 channels and propagating for 20 channels only, then you may use the --spectrum option to set the spectrum you wish to propagate on this already-designed-for-80-channels network. Please bear in mind that GNPy non linear effects computation is accurate for full load evaluation and may loose accuracy with partial loading. The --spectrum option will also enable you to define non-flat spectrum. https://gnpy.readthedocs.io/en/master/json.html#arbitrary-channel-definition.

Regarding your last question about tx_osnr, could you please explain the simulation setup more precisely ? on my side when I set tx_osnr with different values the propagation results change accordingly.

[xiaotianj]

Thank you for the reply.
Currently, I set a list of length 80 (tx_power_W in the following codes) as signal and tx_power in the create_arbitrary_spectral_information function. Is there any difference using the --spectrum option you mentioned? I will try to use --spectrum option to estimate QoT of the same spectrum.

Regarding tx_osnr, I'm using version v2.11 and the following codes for simulation:

equipment = load_equipment(eqpt_library_name)
network = load_network(network_file_name, equipment)
build_network(network, equipment, pref_ch, pref_total)

transceivers = {n.uid: n for n in network.nodes() if isinstance(n, Transceiver)}

si = create_arbitrary_spectral_information(frequency=frequencies, signal=tx_power_W, baud_rate=50e9,
        roll_off=0, tx_osnr=25, tx_power=tx_power_W)
source = next(transceivers[uid] for uid in transceivers if uid == source)#
sink = next(transceivers[uid] for uid in transceivers if uid == dest)
path = dijkstra_path(network, source, sink)

for el in enumerate(path):
    si = el(si)
    print(el)  # remove this line when sweeping across several powers

After running the above code, the following information was printed, the GSNR and OSNR are not correctly:

Transceiver trx 1
  GSNR (0.1nm, dB):          inf
  GSNR (signal bw, dB):      inf
  OSNR ASE (0.1nm, dB):      inf
  OSNR ASE (signal bw, dB):  inf
  CD (ps/nm):                0.00
  PMD (ps):                  0.00
  PDL (dB):                  0.00
  Latency (ms):              0.00
  Actual pch out (dBm):      0.00

The type_variety of transceiver adopted is as follows, it seems that some parameters in "mode" were not used:

{
    "type_variety": "Huawei",
    "frequency": {
        "min": 184e12,
        "max": 197e12
    },
    "mode": [{
            "format": "QPSK",
            "baud_rate": 28e9,
            "OSNR": 25,
            "bit_rate": 100e9,
            "roll_off": 0.0,
            "tx_osnr": 25,
            "min_spacing": 50e9,
            "cost": 1
        }
    ]
}