Resumen
Apparatuses and methods for use by analytics service producers, consumers, storage entities and network exposure entities of a communication network. The method comprises receiving a training request issued by an analytics service consumer of the communication network, the training request comprising an indication of a selected model of a plurality of models, each model of the plurality of models configured to produce analytics for an analytics service; and causing training of the selected model based at least on values of configurable training parameters associated with the selected model to optimize an application metric of the analytics service consumer.
Reivindicaciones
1. A method for use by an analytics service producer of a communication network, the method comprising: receiving a training request issued by an analytics service consumer of the communication network, the training request comprising an indication of a selected model of a plurality of models, each model of the plurality of models configured to produce analytics for an analytics service; and causing training of the selected model based at least on values of configurable training parameters associated with the selected model to optimize an application metric of the analytics service consumer. 2. The method of claim 1, wherein the training request further comprises initial values for one or more of the configurable training parameters associated with the selected model and the method comprises causing setting the values of the one or more configurable training parameters to the initial values. 3. The method of claim 1 or 2, further comprising: causing setting of default values for one or more of the configurable training parameters associated with the selected model. 4. The method of any one of claims 1 to 3, further comprising: providing the analytics service consumer with information about the plurality of models and respective configurable training parameters associated with the plurality of models. 5. The method of any one of claims 1 to 4, further comprising: during training of the selected model, sending status information for the selected model and temporary analytics produced by the selected model towards the analytics service consumer. 6. The method of claim 5, further comprising: receiving updated values for the configurable training parameters from the analytics service consumer, wherein the updated values have been set to optimize the application metric of the analytics service consumer; and causing continuation of the training of the selected model based on the updated values for the configurable training parameters. 7. The method of any one of claims 1 to 6, wherein at least one of the following applies: the analytics service producer comprises the plurality of models; the analytics service producer trains the selected model; the analytics service producer provides access to the analytics service of a network entity of the communication network, which comprises the plurality of models; the analytics service producer provides access to the analytics service of a network entity of the communication network, which trains the selected model; or the analytics service producer provides access to the analytics service of a network entity of the communication network, which comprises the plurality of models and trains the selected model. 8. A method for use by an analytics service consumer of a communication network, the method comprising: selecting a model from a plurality of models for an analytics service, each model of the plurality of models configured to produce analytics for the analytics service; and sending a training request comprising an indication of the model, for training the model based at least on values of configurable training parameters associated with the model, towards an analytics service producer to train the model to optimize an application metric of the analytics service consumer. 9. The method of claim 8, further comprising: setting values for one or more of the configurable training parameters; and including the values in the training request. 10. The method of claim 8 or 9, further comprising at least one of the following: selecting the model from the plurality of models based on information provided by the analytics service producer; selecting the model and the one or more of the configurable training parameters from the plurality of models and configurable training parameters associated with the respective models based on information provided by the analytics service producer; selecting the model from the plurality of models based on information which is provided by a network exposure entity of the communication network based on a profile of the analytics service producer stored in a storage entity of the communication network; or selecting the model and the one or more of the configurable training parameters from the plurality of models and configurable training parameters associated with the respective models based on information which is provided by a network exposure entity of the communication network based on a profile of the analytics service producer stored in a storage entity of the communication network. 11. The method of any one of claims 8 to 10, further comprising: receiving status information of the model being trained and temporary analytics produced by the model being trained; assessing the application metric and determining new values for the configurable training parameters based on the assessing for optimizing the application metric of the analytics service consumer; and updating the values of the configurable training parameters to the new values and sending the updated values towards the analytics service producer. 12. A method for use by a storage entity of a communication network, the method comprising: storing profiles of analytics service producers of network entities of the communication network, the profiles comprising a plurality of models configured to produce analytics for analytics services, and configurable training parameters associated with the plurality of models to be set by analytics service consumers to optimize application metrics of the analytics service consumers; and providing the profiles to a network exposure entity of the communication network. 13. A method for use by a network exposure entity of a communication network, the method comprising: retrieving, based on requested one or more analytics services, profiles of analytics service producers of the communication network, the profiles comprising a plurality of models configured to produce analytics for the requested analytics services, and configurable training parameters associated with the plurality of models to be set by an analytics service consumer to optimize an application metric of the analytics service consumer; and sending the profiles to the analytics service consumer. 14. The method of any one of claims 1 to 13, wherein at least one of the following applies: the analytics service producer comprises at least one of the following: a network data analytics function, an analytics logical function, a model training logical function, a machine learning model training management services producer, a non-real-time radio access network intelligent controller, or a near-real-time radio access network intelligent controller; the analytics service consumer comprises at least one of the following: an application function, a machine learning model training management services consumer, an rApp, or an xApp; the storage entity comprises a network repository function; the network exposure entity comprises a network exposure function; the plurality of models comprises artificial intelligence/machine learning models; the configurable training parameters comprise at least one of the following: network metric training parameters, filters offset of a deformable convolutional layer of a network, or a learning rate applied during the training; the training is performed over a training period corresponding to a defined number of epochs; the analytics comprises prediction; the communication network is a mobile communication network; the network entity comprises a cloud server; or the communication network comprises at least one of the following: a 5G core, or an open radio access network. 15. An apparatus comprising means for: receiving a training request issued by an analytics service consumer of a communication network, the training request comprising an indication of a selected model of a plurality of models, each model of the plurality of models configured to produce analytics for an analytics service; and causing training of the selected model based at least on values of configurable training parameters associated with the selected model to optimize an application metric of the analytics service consumer. 16. An apparatus comprising means for: selecting a model from a plurality of models for an analytics service, each model of the plurality of models configured to produce analytics for the analytics service; and sending a training request comprising an indication of the model, for training the model based at least on values of configurable training parameters associated with the model, towards an analytics service producer to train the model to optimize an application metric of the apparatus. 17. An apparatus comprising means for: storing profiles of analytics service producers of network entities of a communication network, the profiles comprising a plurality of models configured to produce analytics for analytics services, and configurable training parameters associated with the plurality of models to be set by analytics service consumers to optimize application metrics of the analytics service consumers; and providing the profiles to a network exposure entity of the communication network. 18. An apparatus comprising means for: retrieving, based on requested one or more analytics services, profiles of analytics service producers of a communication network, the profiles comprising a plurality of models configured to produce analytics for the requested analytics services, and configurable training parameters associated with the plurality of models to be set by an analytics service consumer to optimize an application metric of the analytics service consumer; and sending the profiles to the analytics service consumer.
