Methods and apparatus relating to machine-learning in a communications network

A method performed by a central Network Data Analytics Function (NWDAF) in a communications network, the method comprising:

• identifying a plurality of local NWDAFs in the communications network that are capable of participating in federated training of a machine-learning model using local data, the machine-learning mode including a plurality of parameters having respective values that are updated based on training, and the identifying based the central NWDAF receiving subscription requests from the plurality of local NWDAFs in the communications network, the subscription requests being requests to participate in federated learning with respect to the machine-learning model;
• receiving from each one among the plurality of local NWDAFs, a respective local model update that contains only locally updated values of the plurality of parameters, as updated by each local NWDAF via local training of the machine-learning model based on respective local data;
• for each parameter among the plurality of parameters, combining the corresponding locally updated values to obtain a respective centrally updated value for the parameter;
• and
• distributing the centrally updated values to the plurality of local NWDAFs on a conditional basis, the condition being more than a threshold change in one or more of the centrally updated values.