Route selection in optical networks based on machine learning
A network node in an optical network dynamically generates a routing table based on attributes of the optical network. The network node obtains attributes characterizing the optical network, which includes multiple network nodes connected by optical links. The network node calculates cost values for sending data from the network node to one or more next hop nodes that are connected to the network node. Each particular cost value is associated with a probability of success of sending the data to a particular next hop node based on a particular permutation of the attributes characterizing the optical network. The network node generates a routing table correlating the permutations of the attributes with each next hop node based on the cost values.
1. A method comprising:
obtaining a plurality of attributes characterizing an optical network, the optical network comprising a plurality of network nodes connected by a plurality of optical links;
at a first network node among the plurality of network nodes, calculating a plurality of cost values for sending data from the first network node to one or more next hop network nodes that are connected to the first network node, wherein a particular cost value among the plurality of cost values is associated with a probability of success of sending the data to a particular next hop network node among the one or more next hop network nodes based on a particular permutation of the plurality of attributes characterizing the optical network;
at the first network node, generating a routing table correlating a plurality of permutations of the attributes with each next hop network node based on the plurality of cost values; and
detecting at least one new network node in the optical network from a new permutation of the attributes characterizing the optical network.
2. The method of claim 1 , wherein the plurality of attributes includes at least one dynamic attribute associated with a recent state of the optical network.
3. The method of claim 1 , wherein the plurality of attributes include an Optical Signal-to-Noise Ratio (OSNR) for optical links between the plurality of network nodes, a fiber cut history for the optical links, a span length of the optical links, and destination identifiers for the network nodes in the optical network.
4. The method of claim 1 , wherein calculating comprises using a machine-learning model on the first network node to calculate the particular cost value for sending data to the particular next hop network node.
5. The method of claim 4 , wherein the machine-learning model is a Bayesian network.
6. The method of claim 4 , wherein the machine-learning model calculates the particular cost value to maximize link utilization across the optical network.
7. The method of claim 4 , wherein the machine-learning model calculates the particular cost value to select paths to minimize an Optical Signal-to-Noise Ratio (OSNR) and a fiber cut history for the optical links.
8. The method of claim 1 , wherein the calculating a respective plurality of cost values and generating a respective routing table is performed at each respective network node of the plurality of network nodes in the optical network.
9. An apparatus comprising:
a network interface configured to communicate with one or more computing device; and
a processor coupled to the network interface, the processor configured to:
obtain via the network interface, a plurality of attributes characterizing an optical network, the optical network comprising a plurality of network nodes including the apparatus;
calculate a plurality of cost values for sending data from the apparatus to one or more next hop network nodes that are connected to the apparatus, wherein a particular cost value among the plurality of cost values is associated with a probability of success of sending the data to a particular next hop network node among the one or more next hop network nodes based on a particular permutation of the plurality of attributes characterizing the optical network;
generate a routing table correlating a plurality of permutations of the attributes with each next hop network node based on the plurality of cost values; and
detect at least one new network node in the optical network from a new permutation of the attributes characterizing the optical network.
10. The apparatus of claim 9 , wherein the plurality of attributes includes at least one dynamic attribute associated with a recent state of the optical network.
11. The apparatus of claim 9 , wherein the plurality of attributes include an Optical Signal-to-Noise Ratio (OSNR) for optical links between the plurality of network nodes, a fiber cut history for the optical links, a span length of the optical links, and destination identifiers for the network nodes in the optical network.
12. The apparatus of claim 9 , wherein the processor is configured to calculate the plurality of cost values by using a machine-learning model on the apparatus to calculate the particular cost value for sending data to the particular next hop network node.
13. The apparatus of claim 12 , wherein the machine-learning model is a Bayesian network.
14. The apparatus of claim 12 , wherein the processor is configured to use the machine-learning model to calculate the particular cost value to maximize link utilization across the optical network.
15. One or more non-transitory computer readable storage media encoded with software comprising computer executable instructions and, when the software is executed on a processor of a first network node of an optical network, operable to cause a processor to:
obtain a plurality of attributes characterizing the optical network, the optical network comprising a plurality of network nodes connected by a plurality of optical links;
calculate a plurality of cost values for sending data from the first network node to one or more next hop network nodes that are connected to the first network node, wherein a particular cost value among the plurality of cost values is associated with a probability of success of sending the data to a particular next hop network node among the one or more next hop network nodes based on a particular permutation of the plurality of attributes characterizing the optical network;
generate a routing table correlating a plurality of permutations of the attributes with each next hop network node based on the plurality of cost values; and
detect at least one new network node in the optical network from a new permutation of the attributes characterizing the optical network.
16. The one or more non-transitory computer readable storage media of claim 15 , wherein the software is further operable to cause the processor to calculate the plurality of cost values by using a machine-learning model on the first network node to calculate the particular cost value for sending data to the particular next hop network node.
17. The one or more non-transitory computer readable storage media of claim 16 , wherein the machine-learning model is a Bayesian network.
18. The one or more non-transitory computer readable storage media of claim 16 , wherein the machine-learning model calculates the particular cost value to maximize link utilization across the optical network.
19. The one or more non-transitory computer readable storage media of claim 15 , wherein the plurality of attributes includes at least one dynamic attribute associated with a recent state of the optical network.
20. The one or more non-transitory computer readable storage media of claim 15 , wherein the plurality of attributes include an Optical Signal-to-Noise Ratio (OSNR) for optical links between the plurality of network nodes, a fiber cut history for the optical links, a span length of the optical links, and destination identifiers for the network nodes in the optical network.