# US 11,115,286 B2 · Automatic discovery of IP-optical links with multi-layer filtering and traffic mapping using neural networks

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 11,115,286 B2
- Document ID: US11115286B2
- Status: Granted
- Kind: B2
- Application number: 16/365,201
- Filed: 2019-03-26
- Published: 2020-10-01
- Granted: 2021-09-07
- Earliest claimed: 2019-03-26
- Assignees: Nokia Solutions and Networks Oy
- Applicants: Nokia Solutions and Networks Oy
- Inventors: Rakesh Kumar; Sumanth MS; Joshi Sravan Kumar K
- Web page: https://patentgrep.com/patent/US11115286B2

## CPC

- H04L 41/12: Discovery or management of network topologies
- G06N 3/045: Combinations of networks
- G06N 3/0464: Convolutional networks [CNN, ConvNet]
- G06N 3/084: Backpropagation, e.g. using gradient descent
- G06N 3/09: Supervised learning
- G06N 20/20: Ensemble learning
- H04J 14/0267: {Optical signaling or routing}
- H04L 41/0873: Checking configuration conflicts between network elements
- H04L 41/0893: Assignment of logical groups to network elements
- H04L 41/16: using machine learning or artificial intelligence
- H04L 45/04: {Interdomain routing, e.g. hierarchical routing}
- H04L 45/64: using an overlay routing layer
- G06N 3/08: Learning methods
- H04L 45/507: {Label distribution}
- H04L 45/62: {Wavelength based (optical switching H04Q11/0062)}
- Y02D 30/00: Reducing energy consumption in communication networks
- H04L: TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION (arrangements common to telegraphic and telephonic communication H04M)
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- H04J: MULTIPLEX COMMUNICATION (transmission in general H04B; peculiar to transmission of digital information H04L5/00; systems for the simultaneous or sequential transmission of more than one television signal H04N7/08; in exchanges H04Q11/00; stereophonic systems H04S)
- Y02D: CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE

## Abstract

A method of identifying IP-optical links in a network having a plurality of nodes, including: grouping network nodes into discovery groups; for each group filtering ports of the nodes in the discovery group; for each group producing class IDs for each filtered port using a machine learning model; for each group matching IP ports to optical ports from the filtered ports using the class IDs for each port to identify IP-optical links; and verifying identified IP-optical links.

## Claims

### Claim 1

A method of identifying IP-optical links in a network having a plurality of nodes, the method comprising:
grouping network nodes into discovery groups;
for each discovery group:
filtering ports of the nodes in the discovery group;
producing class IDs for each filtered port using a machine learning model, wherein the machine learning model matches each of a plurality of Tx ports to a respective Rx port tagged with a unique class ID;
matching IP ports to optical ports from the filtered ports using the class IDs of each port to identify IP-optical links; and
verifying the identified IP-optical links.

### Claim 2 (depends on 1)

The method of claim 1, wherein the grouping of the network nodes into the discovery groups further comprises:
using k-means methods based on the client ports of the optical devices connecting to the IP devices.

### Claim 3 (depends on 1)

The method of claim 1, wherein the grouping of network nodes into the discovery groups further comprises:
receiving a radial distance; and
identifying all the nodes in the network that fall within the received radial distance.

### Claim 4 (depends on 1)

The method of claim 1, wherein the filtering of the ports further comprises:
removing all ports with no traffic.

### Claim 5 (depends on 1)

The method of claim 1, wherein the filtering of the ports further comprises:
removing all IP-IP connected ports in a discovery group which have link layer discovery protocol (LLDP) peer ports.

### Claim 6 (depends on 1)

The method of claim 1, wherein the machine learning model is a convolutional neural network model.

### Claim 7 (depends on 1)

The method of claim 1, wherein the verifying of the identified IP-optical links further comprises:
tracing the IP-optical-IP path; and
comparing the traced IP-optical-IP path with the corresponding LLDP/IGP session between the indirectly connected IP nodes.

### Claim 8 (depends on 1)

The method of claim 1, further comprising:
verifying any manually added IP-Optical links to the SDN controller by tracing the IP-Optical-IP path with the corresponding LLD/IGP session between the indirectly connected IP nodes.

### Claim 9

A method of identifying IP-optical links in a network having a plurality of nodes, the method comprising:
grouping network nodes into discovery groups;
for each discovery group:
filtering ports of the nodes in the discovery group;
producing class IDs for each filtered port using a machine learning model, wherein producing the class IDs for each filtered port includes:
collecting a sequence of time-sampled data from all of the ports and forming an image using the collected data;
assigning different class IDs to all of the Tx ports;
training the machine learning model to learn the class ID for each of the Tx ports using the images formed for the Tx ports; and
inputting the images for all of the Rx ports into the machine learning model to assign a class ID to each of the Rx ports;
matching IP ports to optical ports from the filtered ports using the class IDs of each port to identify IP-optical links; and
verifying the identified IP-optical links.

### Claim 10

A software defined network (SDN) controller for identifying IP-optical links in a network having a plurality of nodes, the SDN controller comprising:
a memory; and
a processor connected to the memory, wherein the processor is configured to:
group network nodes into discovery groups;
for each discovery group:
filter ports of the nodes in the discovery group;
produce class IDs for each filtered port using a machine learning model, wherein the machine learning model matches each of a plurality of Tx ports to a respective Rx port tagged with a unique class ID;
match IP ports to optical ports from the filtered ports using the class IDs of each port to identify IP-optical links; and
verify identified IP-optical links.

### Claim 11 (depends on 10)

The SDN controller of claim 10, wherein the grouping of the network nodes into discovery groups includes using k-means methods based on the client ports of the optical devices connecting to the IP devices.

### Claim 12 (depends on 10)

The SDN controller of claim 10, wherein the grouping of the network nodes into discovery groups includes receiving a radial distance and identifying all the nodes in the network that fall within the received radial distance.

### Claim 13 (depends on 10)

The SDN controller of claim 10, wherein the filtering of the ports includes removing all ports with no traffic.

### Claim 14 (depends on 10)

The SDN controller of claim 10, wherein the filtering of the ports includes removing all IP-IP connected ports in a discovery group which have link layer discovery protocol (LLDP) peer ports.

### Claim 15 (depends on 10)

The SDN controller of claim 10, wherein the machine learning model is a convolutional neural network model.

### Claim 16 (depends on 10)

The SDN controller of claim 10, wherein the verifying of the identified IP-optical links includes tracing the IP-optical-IP path and comparing the traced IP-optical-IP path with the corresponding LLDP/IGP session between the indirectly connected IP nodes.

### Claim 17 (depends on 10)

The SDN controller of claim 10, wherein the processor is further configured to verify any manually added IP-Optical links to the SDN controller by tracing the IP-Optical-IP path with the corresponding LLD/IGP session between the indirectly connected IP nodes.

### Claim 18

A software defined network (SDN) controller for identifying IP-optical links in a network having a plurality of nodes, the SDN controller comprising:
a memory; and
a processor connected to the memory, wherein the processor is configured to:
group network nodes into discovery groups;
for each discovery group:
filter ports of the nodes in the discovery group;
produce class IDs for each filtered port using a machine learning model, wherein producing the class IDs for each filtered port includes:
collecting a sequence of time-sampled data from all of the ports and forming an image using the collected data;
assigning different class IDs to all of the Tx ports;
training the machine learning model to learn the class ID for each of the Tx ports using the images formed for the Tx ports; and
inputting the images for all of the Rx ports into the machine learning model to assign a class ID to each of the Rx ports;
match IP ports to optical ports from the filtered ports using the class IDs of each port to identify IP-optical links; and
verify identified IP-optical links.

### Claim 19

A non-transitory machine-readable storage medium encoded with instructions for execution by a software defined network controller, the non-transitory machine-readable storage medium comprising:
instructions for grouping network nodes into discovery groups;
for each discovery group:
instructions for filtering ports of the nodes in the discovery group;
instructions for producing class IDs for each filtered port using a machine learning model, wherein the machine learning model matches each of a plurality of Tx ports to a respective Rx port tagged with a unique class ID;
instructions for matching IP ports to optical ports from the filtered ports using the class IDs of each port to identify IP-optical links; and
instructions for verifying identified IP-optical links.

### Claim 20 (depends on 19)

The non-transitory machine-readable storage medium of claim 19, wherein the filtering of the ports includes removing all IP-IP connected ports in a discovery group which have link layer discovery protocol (LLDP) peer ports.