US 12,133,095 B2
Machine learning-based approaches for service function chain selection
Faraz Ahmed, Milpitas, CA (US); Lianjie Cao, Milpitas, CA (US); and Puneet Sharma, Milpitas, CA (US)
Assigned to Hewlett Packard Enterprise Development LP, Spring, TX (US)
Filed by HEWLETT PACKARD ENTERPRISE DEVELOPMENT LP, Houston, TX (US)
Filed on Oct. 15, 2021, as Appl. No. 17/503,232.
Prior Publication US 2023/0123074 A1, Apr. 20, 2023
Int. Cl. H04W 24/02 (2009.01); G06N 7/01 (2023.01); G06N 20/00 (2019.01); H04W 4/50 (2018.01); H04W 24/10 (2009.01); H04W 40/12 (2009.01); H04W 48/18 (2009.01)
CPC H04W 24/02 (2013.01) [G06N 7/01 (2023.01); G06N 20/00 (2019.01); H04W 4/50 (2018.02); H04W 24/10 (2013.01); H04W 40/12 (2013.01); H04W 48/18 (2013.01)] 15 Claims
OG exemplary drawing
 
1. A computer-implemented method, comprising:
collecting telemetry data indicative of a set of one or more network performance parameters for a mobile communications network;
selecting a network performance parameter;
training, based on the collected telemetry data, a machine learning model to determine a service function chain (SFC) that optimizes the selected network performance parameter for a given incoming network service request;
receiving a set of incoming network service requests including at least a first network service request and a second network service request;
utilizing the trained machine learning model to identify a set of SFCs that optimizes the selected network performance parameter for the set of incoming network service requests, the set of SFCs including a first SFC that optimizes the selected performance parameter with respect to the first network service request and a second SFC different from the first SFC that optimizes the selected performance parameter with respect to the second network service request;
assigning the set of incoming network service requests to the set of SFCs;
determining a current value of the selected network performance parameter associated with the first SFC handling the first network service request; and
updating a set of historically observed values of the selected network performance parameter by including the current value in the set of historically observed values.