Enhanced uncertainty management for optical communication systems
Margin hedging in optical communication systems allows for overhead within the optical system. A machine learning model can be trained using the output of a physics based simulation of the optical system as well as the features of the optical system. A trained machine learning model can adjust the results of a physics based simulation of an optical network to more accurately match the adjusted simulation results to the “true” performance of the optical network. The margins of the optical communication system can be more tailored to the true performance of a designed or planned optical system.
1. A method for predicting link design margins, the method comprising:
using one or more processors configured to control the steps of receiving a plurality of link features in the form of a plurality of values corresponding to respective variables, each value being from a known range or set of values associated with the corresponding variable, the link features being associated with a physical optical link and forming a signature for the physical optical link;
simulating the physical optical link with a first model based on the plurality of link features to produce a first value;
obtaining empirically one or more performance metrics associated with the physical optical link;
calculating a second value based on the one or more performance metrics;
determining one or more prediction error values based on the first value and the second value, wherein the one or more prediction error values are indicative of the accuracy of the simulation of the first value; and
training or updating a machine learning model using the plurality of link features and at least one of the one or more prediction error values or the one or more performance metrics, wherein the training or updating comprises training or updating the machine learning model to output corrections for application to outputs of the first model.
2. The method of claim 1 further comprising providing a predicted link capacity margin for a new physical optical link based on one or more prediction error values.
3. The method of claim 1 further comprises determining variability of each of the one or more prediction error values for each of the plurality of link features.
4. The method of claim 1 wherein the one or more performance metric comprises a generalized signal to noise ratio.
5. A method for predicting link capacity margins, the method comprising:
using one or more processors configured to control the steps of
receiving a plurality of link features in the form of a plurality of values corresponding to respective variables, each value being from a known range or set of values associated with the corresponding variable, the link features being associated with a physical optical link and forming a signature for the physical optical link;
simulating the physical optical link with a first model based on the plurality of link features to produce a first simulated value;
providing to a trained machine learning model the plurality of link features and the first simulated value; and
outputting by the trained machine learning model corrections for application to outputs of the first model and an error range associated with the first simulated value, wherein the error range is indicative of the accuracy of the simulation of the first value.
6. The method of claim 5 further comprising determining a margin based on the output of the trained machine learning model.
7. The method of claim 5 wherein the error range is based on trained variability values.
8. The method of claim 7 further comprising determining a margin for the physical optical link based on the predicted error range.
9. The method claim 5 further comprising providing as an output a relationship between generalized signal to noise ratio and frequency for at least one set of link features.
10. The method of claim 5 wherein the set of link features is one of a multiplexer, amplifier, or optical fiber.
11. A non-transient computer readable medium containing program instructions, the instructions when executed cause one or more processors to control the steps of:
receiving a plurality of link features in the form of a plurality of values corresponding to respective variables, each value being from a known range or set of values associated with the corresponding variable, the link features being associated with a physical optical link and forming a signature for the physical optical link;
simulating the physical optical link with a first model based on the plurality of link features to produce a first simulated value;
providing to a trained machine learning model the plurality of link features and the first simulated value; and
outputting by the trained machine learning model corrections for application to outputs of the first model and an error range associated with the first simulated value, wherein the error range is indicative of the accuracy of the simulation of the first value.
12. The non-transient computer readable medium of claim 11 further comprising determining a margin based on the output of the trained machine learning model.
13. The non-transient computer readable medium of claim 12 wherein the error range is based on trained variability values.
14. The non-transient computer readable medium of claim 11 further comprising providing as an output a relationship between generalized signal to noise ratio and frequency for at least one set of link features.
15. The non-transient computer readable medium of claim 11 further comprising providing as an output a relationship between generalized signal to noise ratio and frequency for at least one set of link features.
16. The non-transient computer readable medium of claim 11 wherein the set of link features contains at least one of a multiplexer, amplifier, or optical fiber.
17. The non-transient computer readable medium of claim 11 wherein the trained machine learning model is trained on a sets of data, each set of data associated with a physical optical network, each set of data comprising (i) a simulated value for the physical optical network, (ii) a set of link features, and (iii) empirical data from the physical optical network.