-Promote incremental and iterative network design
-Automatic saving of a "network-name_auto_design.json" file
-Results of the autodesign (additional amplifiers, configuration) are
saved to this ..._auto_design.json file
-This file can then be used to run a new simulation just like a normal
json input file
Signed-off-by: Jean-Luc Auge <jeanluc.auge@orange.com>
new field 'out_voa_auto': true/false in eqpt_config json for elligible
amplifiers
if the field is set to true and the power_mode (in
eqpt_config[Spans]) is also set to true, a VOA value is calculated based
on available gain and power margins (EOL) to maximize amplifier gain and
improve its NF
Signed-off-by: Jean-Luc Auge <jeanluc.auge@orange.com>
separate in the code:
* add egress amplifier
* vs setting the amplifier parameters
=> prepare improvments in select_edfa and target_power code
regroup all network optimization operations in network.py (remove from
transmission_main)
Signed-off-by: Jean-Luc Auge <jeanluc.auge@orange.com>
* toggle true/false in eqpt_config json file to allow the use of a given amplifier type in automatic design: eqpt_config[Edfa][allowed_for_design] : true/false
* automatic design is picking the best amplifier (gain, NF constraints)
among the available ones: if toggle to false, the type is deemed non
available. Only alternative before this feature was to remove the
amplifier from eqpt_config library.
* if set to false, the amplifier type can still be input in the network
topology file: it will be recongized.
Signed-off-by: Jean-Luc Auge <jeanluc.auge@orange.com>
* automatic design (when amplifiers are missing from network topology
input) finds the optimum power difference between spans
* The range of this optimum power difference is defined in
eqpt_config[Spans][delta_power_range_db] = [min, max, step]
Signed-off-by: Jean-Luc Auge <jeanluc.auge@orange.com>
=>small span padding now looks at the total loss of all spans spliced
together instead of individual span losses
=>the att_in padding attenuator is placed at the input of the 1st span
(the 1st of the serie of spans spliced together with Fused ne)
squashed with the fix to implement att_in attenuation in Fiber.propagate
Signed-off-by: Jean-Luc Auge <jeanluc.auge@orange.com>
intermediate bug fix to be squashed
Signed-off-by: Jean-Luc Auge <jeanluc.auge@orange.com>
read [Spans][padding] in eqpt_config json file
* add input attenuator on all fibers in the network with loss < padding
* create new Fiber attribute: att_in, which is a fixed attenuator for
padding purposes
* define "padding" : 0 in eqpt_config will disable the feature
effectively
improve fiber splitting to take into account the padding target in the
min fiber length and the target length is set accordingly to this min
length
Signed-off-by: Jean-Luc Auge <jeanluc.auge@orange.com>
read fiber spans default values from eqpt_config.json[Spans]
=>if con_in/con_out is None or not defined in params
Rename connector_loss_in/out to con_in/out for consistency
update parser tests to accept the new con_in/out syntax and the new default
values
Signed-off-by: Jean-Luc Auge <jeanluc.auge@orange.com>
- adding relevant call in path_request_run
- unifying variable name: baudrat -> baud_rate
- correcting test with the new power features
- unifying naming of variables n_ch-> nb_channel, sink -> destination (this
follows the usual naming in yang path computation models)
- bug fix in novel trx_mode_params function , use of the function in
path_requests_un
- TODO : place roll-off in the tsp lib instead of SI
Signed-off-by: EstherLerouzic <esther.lerouzic@orange.com>
bugs from the last refactor merge
- fix#1: update fiber length after split
- fix#2: add egress amp after fiber split
Signed-off-by: Jean-Luc Auge <jeanluc.auge@orange.com>
test verifies that OSNR and SNR computed values are consistant
including connector loss and input power variation
onto 2 example links: 1 and 5 spans
Signed-off-by: EstherLerouzic <esther.lerouzic@orange.com>
- connector losses can be part of the json description or not
in any case a connector loss in and out is now part of the fiber class.
if not present the default value is 0.0 dB
- propagate function in Fiber class now has a first attenuation
propagation step. output conn loos is integrated in the
initial loop function.
Signed-off-by: EstherLerouzic <esther.lerouzic@orange.com>
fix bug in Fused span: adding ingress and egress directions
=> do not share the same NE between 2 directions
Signed-off-by: Jean-Luc Auge <jeanluc.auge@orange.com>
*new Fused spans class to support the connection between 2 fibers
without amplification
*add_egress_amplifier adjust the edfa gain to compensate the loss
of n fused spans
*eqpt not found in eqpt library is caught and stop the code
Signed-off-by: Jean-Luc Auge <jeanluc.auge@orange.com>
code fix for the eqpt sheet reading in xls parser
Signed-off-by: Jean-Luc Auge <jeanluc.auge@orange.com>
amplifier design and eqpt parser code fix
Signed-off-by: Jean-Luc Auge <jeanluc.auge@orange.com>
-read eqpt_config.json with various EDFA and fiber type definitions
-std low gain and medium gain EDFA available
-clearer differentiation between the 2 edfa models:
is the advanced_config_from_json field present or not?
Signed-off-by: Jean-Luc Auge <jeanluc.auge@orange.com>
* Add EDFA unitary pytests
check nf calculation, nf models comparison, ase noise profile
elements.py edfa _gain_profile: add div 0 checks when no ripple
Signed-off-by: Jean-Luc Auge <jeanluc.auge@orange.com>
* support of coronet.conus network and add class __repr__
adaptation of convert.py to new code json formating
add_egress_amplifier, split_fiber and calculate_new_length
subs in network.py
elements.py:
code cleaning
adding new attributes in Edfa class (pin_db, pout_db...)
adding or augmenting class __repr__
Signed-off-by: Jean-Luc Auge <jeanluc.auge@orange.com>
* augment amplifier unitary tests and code cleaning
add new tests in amplifier_test.py for pytest
Edfa class code cleaning in elements.py
recheck code coherence and results
Signed-off-by: Jean-Luc Auge <jeanluc.auge@orange.com>
* ROADM support in network.py, convert.py, coronet json, ROADM class
creation of a ROADM class with 20dB loss
convert.py json parser modification to include roadms
network.py modification to automate amplifier placing after roadm
elements.py inclusion of Roadm class
Signed-off-by: Jean-Luc Auge <jeanluc.auge@orange.com>
* Add Travis-CI configuration for continous integration
* Add GN-Model Documentation and auto-doc integration
* Add GN model to Fiber
* Unit of measure conversion adapted to SI
* JSON file based on Orange operator typical input
Signed-off-by: Jean-Luc Auge <jeanluc.auge@orange.com>
* update of the standalone edfa model
creation of a new amlifier2.py = v2
creation of a json parser build_oa_json.py
the parser takes OA.json as input and newOA.json as output
creation of a pytest verification module amplifier_pytest.py
Signed-off-by: Jean-Luc Auge <jeanluc.auge@orange.com>
* put the code together and transmission example script
-basic dijkstra propagation
-ase noise propagation based on amplifier model
-fake nli noise propagation
-integration of the amplifier model
-interpolation function in the edfa class
-code cleaning and units harmonization
Signed-off-by: Jean-Luc Auge <jeanluc.auge@orange.com>
* mv transmission_main_example and rm _main__
Signed-off-by: Jean-Luc Auge <jeanluc.auge@orange.com>
* 2nd edfa model and build_oa_json file
add a dual coil stages edfa model in case the nf polynomial fit is not known
add a build_oa_json file that convert the input files in
edfa_config.json file and pre-calculate the nf_model nf1, nf2 and
delta_p parameters
adding power violation check and input padding (below minimum gain) in the edfa model
class
Signed-off-by: Jean-Luc Auge <jeanluc.auge@orange.com>
