Query on SI parameters

[SaiBhyri]

Hi,

Have below queries on SI and Span parameters and need help to understand better:

power_dbm – Is it the per channel power coming out of the amplifier or per channel launch power into the fiber?
power_range_db – Is it the variation around power_dbm? For eg: If power_dBm is set to +1dBm and power_range_dB is set to [-1,+1,0.5] then would the simulation be performed for Power values = 0dBm, 0.5dBm, 1dBm, 1.5dBm and 2dBm?
What is delta_power_range_db? Sorry, I couldn’t understand well from the documentation.
There is reference to delta_p in above parameter definitions but couldn’t find its definition. What does it signify?
What is target_extended_gain? Its present in json definition but not in documentation.

baud_rate and spacing are present both under SI parameters and Transceiver parameters. How and where are they used?
I have noticed from my simulations that GSNR is independent of baud_rate and spacing defined under SI parameters. Is GSNR completely governed by baud_rate and spacing defined under Transceiver parameters?

Thanks,
Sai

[EstherLerouzic]

Dear SaiBhyri

SI contains the reference channel type used for network design (if no settings are set on amplifiers)

• gnpy-transmission-main also uses this reference channel for propagation simulation
• gnpy-path-request uses the transceiver's mode definition for propagation. So, in this case, the propagation is performed for channels according to Transceiver's mode. Performance for each service is computed on a worst case basis, assuming the whole spectrum is filled with this transceiver type_variety and mode.

If no settings are set on (some or all) amplifiers the simulator automatically selects one amplifier type_variety among the ones defined in the library that can support the total power target and gain. The total power target is computed for the target amount of channels of the reference type defined in SI. So each amplifier must support

• (f_max - f_min)/(spacing) nb of channels,
• each with power_dbm target power out.

power_dbm is the target power per channel at the output of amplifiers. If a fiber is connected to this amp, then this corresponds to the per channel target power at the input of the fiber span = input power per span.

The optimum input power per span is not the same for all spans (depends on length and loss), so delta_power_range_db enables a range around power_dbm to optimize the input power per span. This deviation around power_dbm is recorded into delta_p in each amplifier instance in the topology. When set to [0, 0, xxx] no optimization is done, so the design strictly uses power_dbm. If the amplifier instance already contains a setting, no computation or optimization are applied and the user settings are applied unless they are not consistent (for example in case settings are not consistent with max power of amplifier type_variety).

If the amplifier maximum power can not support the computed amount of channels with this target power, a reduction is applied and recorded into delta_p.

power_range_db exactly does what you say, a power sweep: (after performing a design for the target channel with power power_dbm, if no settings are present in one or some amplifiers), it propagates a signal for the same type of channel but different target powers. Computed gains remain the ones of the design unless the new computed target_power per channel hits the max power of this amplifier type_variety. This feature enables to explore different target power and identify the best one (to maximize GSNR for example). In case amplifiers had non zero delta_p defined in their operationnal settings, these are also applied so that the whole link target powers are translated up or down depending on the range.

Hope this helps,

regards

[SaiBhyri]

Thanks Esther for detailed response.

[SaiBhyri]

Hi Esther,

Need your help on further queries:

• If there is more than one amplifier satisfying total power target and gain, do we choose the one which gives best OSNR based on NF?

• While running gnpy-path-request and in the case of having more than one transceiver type, is the per channel launch power into the span the worst one among the per channel span launch power calculated for every transceiver mode?

• Similarly, for the above case, is the amplifier selection based on worst/most demanding total power target and gain (as total no. of channels will differ based on transceiver mode definition) among all the transceiver modes?

• Is the optimization of per span launch power done of get best GSNR from every ROADM-ROADM section? For eg: Consider a linear network : Site 1 (ROADM) <-> Site 2 (ILA) <-> Site 3 (ROADM) <-> Site 4 (ILA) <-> Site 5 (ROADM). There are 4 fiber spans and 2 ROADM-ROADM sections. Is the per span launch power optimization done independently on both ROADM-ROADM section?

• Is the per span launch power optimization logic following (considering above topology and with power_dBm = 0dBm and power_range_db = [-1,+1,0.5] ).
  - Site 1 BST power vary from -1dBm to +1dBm in steps of 0.5dB keeping the BST output powers at Site2, Site 3 at 0dBm.
  - Choose the output power value at Site 1 BST which gives the best GSNR from above run.
  - Lock the Site 1 BST value to above found value. Vary Site 2 BST output power from -1dBm to +1dBm in steps of 0.5dB and
  keep BST output power at Site3 at 0dBm.
  - Choose the output power value at Site 2 BST which gives the best GSNR from above run.
  - Lock the Site 1 BST and Site 2 BST value to above found values. Vary Site 3 BST output power from -1dBm to +1dBm in steps
  of 0.5dB.
  - In this manner, find the optimal launch power values for BSTs at Site 1, Site 2, Site 3.
  - Now come back to Site 1 BST and vary its power value from -1dBm to +1dBm in steps of 0.5dB keeping the BST amps output
  power Site 2, Site 3 at their optimal power value found from above runs.
  - If Site 1 BST gives same output power value as before, then move to Site 2, Site 3 and check the same.
  - If above run gives the same BST out power values at Site 1, Site 2,Site 3 then convergence has happened else repeat all the
  above runs till a convergence is reached

Thanks,
Sai

[SaiBhyri]

Hi Esther,

Could you please help on my further queries?

Thanks,
Sai

[EstherLerouzic]

Dear sai,

here are some answers:

• If there is more than one amplifier satisfying total power target and gain, do we choose the one which gives best OSNR based on NF?

yes, it selects the one with lowest NF

• While running gnpy-path-request and in the case of having more than one transceiver type, is the per channel launch power into the span the worst one among the per channel span launch power calculated for every transceiver mode?

No, each request performance is computed independantly from the other request: the propagation is performed full load (according to request setting) for the same channel (all channels with the same power, baudrate, spacing, ... settings). If you want to simulate mixed channel, you must use transmission program with the -spectrum option.

• Similarly, for the above case, is the amplifier selection based on worst/most demanding total power target and gain (as total no. of channels will differ based on transceiver mode definition) among all the transceiver modes?

Selection of amplifier is performed base on the reference channel defined in SI part of eqpt_config.json input. Then propagation for each request uses this very first design. so the amplifier type do not depend on the requests. The gain target or delta_p may however be reduced if the request specifies the channel type and number that leads to a total power that exceeds chosen amp capability (saturation).

• Is the optimization of per span launch power done of get best GSNR from every ROADM-ROADM section? For eg: Consider a linear network : Site 1 (ROADM) <-> Site 2 (ILA) <-> Site 3 (ROADM) <-> Site 4 (ILA) <-> Site 5 (ROADM). There are 4 fiber spans and 2 ROADM-ROADM sections. Is the per span launch power optimization done independently on both ROADM-ROADM section?

optimum channel power is performed per OMS. so in your example Site2 settings is independant from Site1 settings.

• Is the per span launch power optimization logic following (considering above topology and with power_dBm = 0dBm and power_range_db = [-1,+1,0.5] ).
  • Site 1 BST power vary from -1dBm to +1dBm in steps of 0.5dB keeping the BST output powers at Site2, Site 3 at 0dBm.
  • Choose the output power value at Site 1 BST which gives the best GSNR from above run.
  • Lock the Site 1 BST value to above found value. Vary Site 2 BST output power from -1dBm to +1dBm in steps of 0.5dB and
    keep BST output power at Site3 at 0dBm.
  • Choose the output power value at Site 2 BST which gives the best GSNR from above run.
  • Lock the Site 1 BST and Site 2 BST value to above found values. Vary Site 3 BST output power from -1dBm to +1dBm in steps
    of 0.5dB.
  • In this manner, find the optimal launch power values for BSTs at Site 1, Site 2, Site 3.
  • Now come back to Site 1 BST and vary its power value from -1dBm to +1dBm in steps of 0.5dB keeping the BST amps output
    power Site 2, Site 3 at their optimal power value found from above runs.
  • If Site 1 BST gives same output power value as before, then move to Site 2, Site 3 and check the same.
  • If above run gives the same BST out power values at Site 1, Site 2,Site 3 then convergence has happened else repeat all the
    above runs till a convergence is reached

if you are talking of delta_ power_range_db defined in Span:
No each span has its own optimum power accounting for the already propagated power within the OMS. There is no global optimization with a feedback loop as you suggest. It is just one iteration.
So,

• Site 1 BST power varies from -1dBm to +1dBm in steps of 0.5dB (delta_power_range_db) (compute the optimum from the simplified equation in target_power function in network.py). input power for the amp is the one set out of the Site1 ROADM (target_pch_out_db)
• Site 2 ILA power varies from -1dBm to +1dBm in steps of 0.5dB. input power for the amp is the result of Site1 booster gain (and delta_p and voa) setting computed in previous step.
• Site3 preamp power varies from -1dBm to +1dBm in steps of 0.5dB. input power for the amp is the result of Site2 ILA gain setting computed in previous step.
• Site3 booster power varies from -1dBm to +1dBm in steps of 0.5dB. input power for the amp is the one set out of the Site2 ROADM.
• ...
  Please have a look to this paper , this clarifies what's in the current code J. -L. Auge, V. Curri and E. Le Rouzic, Open Design for Multi-Vendor Optical Networks, OFC 2019, equation 4. See delta_power_range_db for more explaination.

power_range_db (from SI) is no used with path-request-run.

hope this helps

regards

[SaiBhyri]

Thanks Esther.

Regards,
Sai