AI-assisted WAN link selection for SD-WAN services

Patent metadata

Field	Value
Publication number	US-12476918-B2
Application number	US-202418628122-A
Country	US
Kind code	B2
Filing date	Apr 5, 2024
Priority date	Sep 30, 2021
Publication date	Nov 18, 2025
Grant date	Nov 18, 2025

Abstract

An example method includes receiving, by a software-defined networking in a wide area network (SD-WAN) system having a first WAN link and a second WAN link for an SD-WAN service, WAN link characterization data for the first WAN link over a time period; determining, by the SD-WAN system based on processing the WAN link characterization data for the first WAN link using a machine learning model trained with historical WAN link characterization data for one or more WAN links, an indicator of a predicted performance metric of the first WAN link at a future time; and reassigning, by the SD-WAN system based on the indicator, an application from the first WAN link to the second WAN link.

First claim

What is claimed is:

1 . A Software-defined (SD) Wide Area Network (WAN) (SD-WAN) edge device comprising:
processing circuitry; and memory storing instructions, that when executed, cause the processing circuitry to:
obtain link characterization data for a first WAN link communicatively coupled to the SD-WAN edge device over a time period,
determine, based on processing the link characterization data for the first WAN link using a machine learning model trained with historical characterization data for one or more WAN links, an indicator of a performance metric of the first WAN link at a future time and a time interval associated with the indicator of the performance metric, and
assign, based on a determination of whether the time interval is predicted to begin after or before a session associated with an application assigned to the first WAN link is predicted to end, the application to remain on the first WAN link or move to a second WAN link communicatively coupled to the SD-WAN edge device instead of the first WAN link, respectively.

2 . The SD-WAN edge device of claim 1 ,
wherein the instructions to assign the application comprise instructions to assign, further based on a determination that the time interval associated with the indicator of the performance metric exceeds a tolerance interval associated with the application, the application to the second WAN link instead of the first WAN link.

3 . The SD-WAN edge device of claim 1 , wherein the link characterization data comprises one or more of service data for the first WAN link, performance metric data indicating measured values for the first WAN link over the time period, a time to first packet, an average length of sessions, or a packet retransmission rate.

4 . The SD-WAN edge device of claim 1 , wherein the instructions further cause the processing circuitry to:
determine the time interval is predicted to begin after the session is predicted to end, and wherein the processing circuitry being configured to assign the application comprises the processing circuitry being configured to assign the application to remain on the first WAN link.

5 . The SD-WAN edge device of claim 1 , wherein the instructions further cause the processing circuitry to:
determine the time interval is predicted to begin before the session is predicted to end, and wherein the processing circuitry being configured to assign the application comprises the processing circuitry being configured to assign the application to the second WAN link instead of the first WAN link.

6 . The SD-WAN edge device of claim 1 , wherein a session interval for the application is a length of time the session associated with the application is expected to last,
wherein the session interval for the application comprises a first session interval, wherein the application comprises a first application assigned to the first WAN link, and wherein the first session interval has a value different than that of a second session interval of a second application assigned to the first WAN link, wherein the instructions that cause the processing circuitry to assign the application comprises instructions that cause the processing circuitry to assign the first application to the second WAN link instead of the first WAN link, wherein the instructions further cause the processing circuitry to: assign, based on the time interval and the second session interval for the second application, the second application to remain on the first WAN link.

7 . The SD-WAN edge device of claim 1 , wherein the SD-WAN edge device is positioned at an edge of a WAN that includes one or more intermediate routers, wherein instructions that cause the processing circuitry to obtain the link characterization data for the first WAN link comprises instructions that cause the processing circuitry to obtain the WAN link characterization data from the one or more intermediate routers.

8 . The SD-WAN edge device of claim 1 , wherein the SD-WAN edge device is positioned at an edge of a WAN that includes one or more intermediate routers and a plurality of other SD-WAN edge devices, wherein instructions that cause the processing circuitry to obtain the link characterization data for the first WAN link comprises instructions that cause the processing circuitry to obtain the link characterization data from an SD-WAN system that is in communication with the one or more intermediate routers and the plurality of other SD-WAN edge devices.

9 . The SD-WAN edge device of claim 1 , wherein the instructions further cause the processing circuitry to determine when the session associated with the application is predicted to end based on at least one of an average session length for the application, a minimum session length for the application, or and a maximum session length for the application.

10 . The SD-WAN edge device of claim 1 ,
wherein the indicator of the performance metric of the first WAN link indicates a predicted interval of network instability of the first WAN link is expected to begin at the future time, and wherein the time interval associated with the indicator comprises an expected duration of the predicted interval of network instability.

11 . The SD-WAN edge device of claim 10 , wherein the instructions further cause the processing circuitry to determine whether the session associated with the application will end prior to the future time at which the predicted interval of network instability of the first WAN link is expected to begin.

12 . The SD-WAN edge device of claim 1 , wherein the indicator of the performance metric of the first WAN link at the future time comprises a vector of predicted future values of performance metrics.

13 . A method comprising:
obtaining, by a Software-defined (SD) Wide Area Network (WAN) (SD-WAN) edge device, link characterization data for a first WAN link communicatively coupled to the SD-WAN edge device over a time period, determining, by the SD-WAN edge device and based on processing the link characterization data for the first WAN link using a machine learning model trained with historical characterization data for one or more WAN links, an indicator of a performance metric of the first WAN link at a future time and a time interval associated with the indicator of the performance metric, and assigning, by the SD-WAN edge device and based on a determination of whether the time interval is predicted to begin after or before a session associated with an application assigned to the first WAN link is predicted to end, the application to remain on the first WAN link or move to a second WAN link communicatively coupled to the SD-WAN edge device instead of the first WAN link, respectively.

14 . The method of claim 13 , wherein the indicator of the performance metric of the first WAN link at the future time comprises a vector of predicted future values of performance metrics.

15 . Non-transitory computer-readable storage media comprising instructions that, when executed by one or more processors of a Software-defined (SD) Wide Area Network (WAN) (SD-WAN) edge device, cause the one or more processors to:
obtain link characterization data for a first WAN link communicatively coupled to the SD-WAN edge device over a time period, determine, based on processing the link characterization data for the first WAN link using a machine learning model trained with historical characterization data for one or more WAN links, an indicator of a performance metric of the first WAN link at a future time and a time interval associated with the indicator of the performance metric, and assign, based on

Assignees

• Juniper Networks Inc

Inventors

• Wang Jisheng
• Baj Michael
• Timmons Patrick
• Melampy Patrick J
• Agrawal Kaushik Adesh
• Mehta Ruchit Rajkumar

Classifications

• H04L41/16

  using machine learning or artificial intelligence · CPC title

• H04L47/2425 Primary

  for supporting services specification, e.g. SLA · CPC title

• H04L67/02 Primary

  based on web technology, e.g. hypertext transfer protocol [HTTP] · CPC title