User plane function (UPF) selection based on predicted load information

A network node comprising:

• one or more processors;
• one or more interfaces configured to connect in a mobile network;
• the one or more processors of the network node having a network data analytics function being configured for operation including:
• receiving a message which indicates a request for one or more predicted load contribution factors indicative of a predicted load contribution associated with a protocol data unit (PDU) session to be established for one or more user equipments (UEs);
• calculating the one or more predicted load contribution factors with use of a model which is derived based on historical usage data of previous PDU sessions of the one or more UEs, wherein the model is a linear regression model that is based on session-related parameters of the PDU session to be established for the one or more UEs and the historical usage data, wherein the historical usage data used to derive the model are obtained from charging data records (CDRs) of the previous PDU sessions of the one or more UEs, and wherein the session-related parameters comprise:
• a day parameter, a time parameter, a device type parameter, a radio access technology (RAT) type parameter, and a subscription parameter;
• regularly rebuilding the model based on feedback data obtained from further CDRs of further PDU sessions of the one or more UEs;
• and
• sending a message which includes the one or more predicted load contribution factors indicative of the predicted load contribution associated with the PDU session to be established for the one or more UEs.

A computer program product comprising:

• a non-transitory computer readable medium;
• instructions stored in the non-transitory computer readable medium, the instructions being executable by one of more processors of a network node for operation as a network data analytics function including for:
• receiving a message which indicates a request for one or more predicted load contribution factors indicative of a predicted load contribution at a user plane function associated with a protocol data unit (PDU) session to be established for one or more user equipments (UEs);
• calculating the one or more predicted load contribution factors with use of a model which is derived based on historical usage data of previous PDU sessions of the one or more UEs, wherein the model is a linear regression model that is based on session-related parameters of the PDU session to be established for the one or more UEs and the historical usage data, wherein the historical usage data used to derive the model are obtained from charging data records (CDRs) of the previous PDU sessions of the one or more UEs, and wherein the session-related parameters comprise:
• a day parameter, a time parameter, a device type parameter, a radio access technology (RAT) type parameter, and a subscription parameter;
• regularly rebuilding the model based on feedback data obtained from further CDRs of further PDU sessions of the one or more UEs;
• and
• sending a message which includes the one or more predicted load contribution factors indicative of the predicted load contribution at the user plane function associated with the PDU session to be established for the one or more UEs.