WO2023036436A1

WO2023036436A1 - Apparatus, methods, and computer programs

Info

Publication number: WO2023036436A1
Application number: PCT/EP2021/074988
Authority: WO
Inventors: Saurabh Khare; Dario BEGA; Muhammad Majid BUTT; Gerald KUNZMANN; Chaitanya Aggarwal
Original assignee: Nokia Technologies Oy
Priority date: 2021-09-10
Filing date: 2021-09-10
Publication date: 2023-03-16

Abstract

There is provided a method, computer program and apparatus for a network analytics function that causes the network analytics functions to: cause a first model representation to be stored, the first model representation being configured to produce a first analytics output having a first quality when executed; cause a second model representation to be stored, the second model representation being configured to produce a second analytics output having a second quality when executed, the first and second model representations being different manifestations of a same model and the second quality being higher than the first quality; receive, from a user equipment, a request for the second analytics output; and signal, to the user equipment, at least one of said second analytics output and/or indication of a location of the second model representation.

Description

APPARATUS, METHODS, AND COMPUTER PROGRAMS

Field

[0001]The present disclosure relates to apparatus, methods, and computer programs, and in particular but not exclusively to apparatus, methods and computer programs for network apparatuses.

Background

[0002]A communication system can be seen as a facility that enables communication sessions between two or more entities such as user terminals, access nodes and/or other nodes by providing carriers between the various entities involved in the communications path. A communication system can be provided for example by means of a communication network and one or more compatible communication devices. The communication sessions may comprise, for example, communication of data for carrying communications such as voice, electronic mail (email), text message, multimedia and/or content data and so on. Content may be multicast or uni-cast to communication devices.

[0003]A user can access the communication system by means of an appropriate communication device or terminal. A communication device of a user is often referred to as user equipment (UE) or user device. The communication device may access a carrier provided by an access node and transmit and/or receive communications on the carrier.

[0004]The communication system and associated devices typically operate in accordance with a required standard or specification which sets out what the various entities associated with the system are permitted to do and how that should be achieved. Communication protocols and/or parameters which shall be used for the connection are also typically defined. One example of a communications system is UTRAN (3G radio). Another example of an architecture that is known is the long-term evolution (LTE) or the Universal Mobile Telecommunications System (UMTS) radioaccess technology. Another example communication system is so called 5G system that allows user equipment (UE) or user device to contact a 5G core via e.g. new radio (NR) access technology or via other access technology such as Untrusted access to 5GC or wireline access technology.

Summary

[0005]According to a first aspect, there is provided an apparatus for a network analytics function, the apparatus comprising means for: causing a first model representation to be stored, the first model representation being configured to produce a first analytics output having a first quality when executed; causing a second model representation to be stored, the second model representation being configured to produce a second analytics output having a second quality when executed, the first and second model representations being different manifestations of a same model and the second quality being higher than the first quality; receiving, from a user equipment, a request for the second analytics output; and signalling, to the user equipment, at least one of said second analytics output and/or indication of a location of the second model representation.

[0006] The apparatus may comprise means for generating the first and second model representations from the same model.

[0007]The apparatus may comprise means for: receiving, from a network function or directly from a user equipment, a request for a user equipment to receive said same model; and signalling metadata identifying at least the first model representation to the network function or the user equipment.

[0008] Said means for signalling metadata may further comprise means for signalling metadata identifying a third model representation configured to produce a third analytics output having a third quality when executed, the third model representation being a manifestation of the same model, and the third accuracy having a value between a value for the first accuracy and a value for the second accuracy.

[0009] The apparatus may comprise means for selecting the first model representation in dependence on at least one of a processing capability of the user equipment and/or a network capacity for downloading the first model representation to the user equipment.

[0010] The apparatus may comprise means for: receiving a request to execute the second model representation; and executing the second model representation using the first analytics output as an input to the second model representation when the request comprises the first analytics output, and/or executing the second model representation using input data that is common to both the first and second model representations.

[0011]The request for the second analytics output may comprise said metadata. The request for the second analytics output may comprise the first analytics output.

[0012] The metadata may comprise an address, and wherein said downloading may comprise downloading the first model representation from said address.

[0013] The metadata may comprise at least one of: an indication of how resource intensive at least one of the model representations is to execute relative to other model representations of said model; an indication of an address from which the first model representation may be downloaded; an indication of a version of the first model representation to be executed; an indication of a version of said model; an indication of a type of said model; an indication of an address from which a second model representation may be downloaded, the second model representation being another manifestation of said model and having a second analytics output associated with a second accuracy, the second accuracy being higher than the first accuracy; an indication of an address from which a third model representation may be downloaded, the third model representation being another manifestation of said model and having a third analytics output associated with a third accuracy, the third accuracy being between the first accuracy and the second accuracy; an indication of an address from which a fourth model representation may be downloaded, the fourth model representation being another manifestation of said model and having a fourth analytics output associated with a fourth accuracy, the fourth accuracy being higher than the second accuracy; and metadata related to analytics output(s) of at least one of the first to fourth model representations.

[0014] The network analytics function may comprise at least one of a network data analytics function, a management data analytics service, and/or any network function or network application function configured to perform analytics.

[0015] According to a second aspect, there is provided an apparatus for a user equipment, the apparatus comprising means for: executing a first model representation to produce a first analytics output having a first quality; determining that a second analytics output having a second quality, higher than the first quality, is desired; signalling, to a network analytics function, a request for the second analytics output; and receiving, from the network analytics function, at least one of said second analytics output and/or indication of a location of a second model representation for obtaining the second analytics output.

[0016] The apparatus may comprise means for: signalling, to a network function and/or to the network analytics function, a request for a user equipment to receive a model for obtaining an analytics output; receiving, from the network function and/or from the network analytics function, metadata identifying at least the first model representation in response to the signalled request; and downloading the first model representation in response to receiving the metadata.

[0017] The request for the second analytics output may comprise said metadata. The request for the second analytics output may comprise the first analytics output.

[0018] The metadata may comprise an address, and wherein said downloading may comprise downloading the first model representation from said address.

[0019] The apparatus may comprise means for receiving, from the network function and/or from the network analytics function, metadata identifying a third model representation configured to produce a third analytics output having a third quality when executed, the third model representation being a manifestation of said model, and the third accuracy having a value between a value for the first accuracy and a value for the second accuracy.

[0020] The apparatus may comprise means for signalling a capability of the user equipment to download and/or execute a model to a network function.

[0021]The capability may be communicated in a registration message.

[0022] The metadata may comprise at least one of: an indication of how resource intensive at least one of the model representations is to execute relative to other model representations of said model; an indication of an address from which the first model representation may be downloaded; an indication of a version of the first model representation to be executed; an indication of a version of said model; an indication of a type of said model; an indication of an address from which a second model representation may be downloaded, the second model representation being another manifestation of said model and having a second analytics output associated with a second accuracy, the second accuracy being higher than the first accuracy; an indication of an address from which a third model representation may be downloaded, the third model representation being another manifestation of said model and having a third analytics output associated with a third accuracy, the third accuracy being between the first accuracy and the second accuracy; an indication of an address from which a fourth model representation may be downloaded, the fourth model representation being another manifestation of said model and having a fourth analytics output associated with a fourth accuracy, the fourth accuracy being higher than the second accuracy; and metadata related to analytics output(s) of at least one of the first to fourth model representations.

[0023] The network analytics function may comprise at least one of a network data analytics function, a management data analytics service, and/or any network function or network application function configured to perform analytics.

[0024]According to a third aspect, there is provided an apparatus for a network function, the apparatus comprising means for: receiving, from a user equipment, a request for the user equipment to receive a model for obtaining an analytics output; signalling the received request to a network analytics function; receiving, from the network analytics function, metadata identifying a first model representation in response to the signalled request, wherein the first model is a manifestation of said model; and signalling the received metadata to the user equipment.

[0025] The apparatus may comprise means for: receiving, from the user equipment, a capability of the user equipment to download and/or execute said model; and signalling the capability to another network function.

[0026]The capability may be communicated in a registration message.

[0027] The metadata may comprise at least one of: an indication of how resource intensive at least one of the model representations is to execute relative to other model representations of said model; an indication of an address from which the first model representation may be downloaded; an indication of a version of the first model representation to be executed; an indication of a version of said model; an indication of a type of said model; an indication of an address from which a second model representation may be downloaded, the second model representation being another manifestation of said model and having a second analytics output associated with a second accuracy, the second accuracy being higher than the first accuracy; an indication of an address from which a third model representation may be downloaded, the third model representation being another manifestation of said model and having a third analytics output associated with a third accuracy, the third accuracy being between the first accuracy and the second accuracy; an indication of an address from which a fourth model representation may be downloaded, the fourth model representation being another manifestation of said model and having a fourth analytics output associated with a fourth accuracy, the fourth accuracy being higher than the second accuracy; and metadata related to analytics output(s) of at least one of the first to fourth model representations.

[0028] The network analytics function may comprise at least one of a network data analytics function, a management data analytics service, and/or any network function or network application function configured to perform analytics.

[0029]According to a fourth aspect, there is provided an apparatus for a network analytics function, the apparatus comprising: at least one processor; and at least one memory comprising code that, when executed by the at least one processor, causes the apparatus to: cause a first model representation to be stored, the first model representation being configured to produce a first analytics output having a first quality when executed; cause a second model representation to be stored, the second model representation being configured to produce a second analytics output having a second quality when executed, the first and second model representations being different manifestations of a same model and the second quality being higher than the first quality; receive, from a user equipment, a request for the second analytics output; and signal, to the user equipment, at least one of said second analytics output and/or indication of a location of the second model representation.

[0030] The apparatus may be caused to generate the first and second model representations from the same model.

[0031]The apparatus may be caused to: receive, from a network function or directly from a user equipment, a request for a user equipment to receive said same model; and signal metadata identifying at least the first model representation to the network function or the user equipment.

[0032] Said signalling metadata may comprise signalling metadata identifying a third model representation configured to produce a third analytics output having a third quality when executed, the third model representation being a manifestation of the same model, and the third accuracy having a value between a value for the first accuracy and a value for the second accuracy.

[0033] The apparatus may be caused to select the first model representation in dependence on at least one of a processing capability of the user equipment and/or a network capacity for downloading the first model representation to the user equipment. [0034] The apparatus may be caused to: receive a request to execute the second model representation; and execute the second model representation using the first analytics output as an input to the second model representation when the request comprises the first analytics output, and/or execute the second model representation using input data that is common to both the first and second model representations.

[0035]The request for the second analytics output may comprise said metadata. The request for the second analytics output may comprise the first analytics output.

[0036] The metadata may comprise an address, and wherein said downloading may comprise downloading the first model representation from said address.

[0037] The metadata may comprise at least one of: an indication of how resource intensive at least one of the model representations is to execute relative to other model representations of said model; an indication of an address from which the first model representation may be downloaded; an indication of a version of the first model representation to be executed; an indication of a version of said model; an indication of a type of said model; an indication of an address from which a second model representation may be downloaded, the second model representation being another manifestation of said model and having a second analytics output associated with a second accuracy, the second accuracy being higher than the first accuracy; an indication of an address from which a third model representation may be downloaded, the third model representation being another manifestation of said model and having a third analytics output associated with a third accuracy, the third accuracy being between the first accuracy and the second accuracy; an indication of an address from which a fourth model representation may be downloaded, the fourth model representation being another manifestation of said model and having a fourth analytics output associated with a fourth accuracy, the fourth accuracy being higher than the second accuracy; and metadata related to analytics output(s) of at least one of the first to fourth model representations.

[0038] The network analytics function may comprise at least one of a network data analytics function, a management data analytics service, and/or any network function or network application function configured to perform analytics.

[0039] According to a fifth aspect, there is provided an apparatus for a user equipment, the apparatus comprising: at least one processor; and at least one memory comprising code that, when executed by the at least one processor, causes the apparatus to: execute a first model representation to produce a first analytics output having a first quality; determine that a second analytics output having a second quality, higher than the first quality, is desired; signal, to a network analytics function, a request for the second analytics output; and receive, from the network analytics function, at least one of said second analytics output and/or indication of a location of a second model representation for obtaining the second analytics output.

[0040] The apparatus may be caused to: signal, to a network function and/or to the network analytics function, a request for a user equipment to receive a model for obtaining an analytics output; receive, from the network function and/or from the network analytics function, metadata identifying at least the first model representation in response to the signalled request; and download the first model representation in response to receiving the metadata.

[0041]The request for the second analytics output may comprise said metadata. The request for the second analytics output may comprise the first analytics output.

[0042] The metadata may comprise an address, and wherein said downloading may comprise downloading the first model representation from said address.

[0043] The apparatus may be caused to receive, from the network function and/or from the network analytics function, metadata identifying a third model representation configured to produce a third analytics output having a third quality when executed, the third model representation being a manifestation of said model, and the third accuracy having a value between a value for the first accuracy and a value for the second accuracy.

[0044] The apparatus may be caused to signal a capability of the user equipment to download and/or execute a model to a network function.

[0045]The capability may be communicated in a registration message.

[0046] The metadata may comprise at least one of: an indication of how resource intensive at least one of the model representations is to execute relative to other model representations of said model; an indication of an address from which the first model representation may be downloaded; an indication of a version of the first model representation to be executed; an indication of a version of said model; an indication of a type of said model; an indication of an address from which a second model representation may be downloaded, the second model representation being another manifestation of said model and having a second analytics output associated with a second accuracy, the second accuracy being higher than the first accuracy; an indication of an address from which a third model representation may be downloaded, the third model representation being another manifestation of said model and having a third analytics output associated with a third accuracy, the third accuracy being between the first accuracy and the second accuracy; an indication of an address from which a fourth model representation may be downloaded, the fourth model representation being another manifestation of said model and having a fourth analytics output associated with a fourth accuracy, the fourth accuracy being higher than the second accuracy; and metadata related to analytics output(s) of at least one of the first to fourth model representations.

[0047] The network analytics function may comprise at least one of a network data analytics function, a management data analytics service, and/or any network function or network application function configured to perform analytics.

[0048]According to a sixth aspect, there is provided an apparatus for a network function, the apparatus comprising: at least one processor; and at least one memory comprising code that, when executed by the at least one processor, causes the apparatus to: receive, from a user equipment, a request for the user equipment to receive a model for obtaining an analytics output; signal the received request to a network analytics function; receive, from the network analytics function, metadata identifying a first model representation in response to the signalled request, wherein the first model is a manifestation of said model; and signal the received metadata to the user equipment.

[0049] The apparatus may be caused to: receive, from the user equipment, a capability of the user equipment to download and/or execute said model; and signal the capability to another network function.

[0050]The capability may be communicated in a registration message.

[0051]The metadata may comprise at least one of: an indication of how resource intensive at least one of the model representations is to execute relative to other model representations of said model; an indication of an address from which the first model representation may be downloaded; an indication of a version of the first model representation to be executed; an indication of a version of said model; an indication of a type of said model; an indication of an address from which a second model representation may be downloaded, the second model representation being another manifestation of said model and having a second analytics output associated with a second accuracy, the second accuracy being higher than the first accuracy; an indication of an address from which a third model representation may be downloaded, the third model representation being another manifestation of said model and having a third analytics output associated with a third accuracy, the third accuracy being between the first accuracy and the second accuracy; an indication of an address from which a fourth model representation may be downloaded, the fourth model representation being another manifestation of said model and having a fourth analytics output associated with a fourth accuracy, the fourth accuracy being higher than the second accuracy; and metadata related to analytics output(s) of at least one of the first to fourth model representations.

[0052] The network analytics function may comprise at least one of a network data analytics function, a management data analytics service, and/or any network function or network application function configured to perform analytics.

[0053]According to a seventh aspect, there is provided a method for an apparatus for a network analytics function, the method comprising: causing a first model representation to be stored, the first model representation being configured to produce a first analytics output having a first quality when executed; causing a second model representation to be stored, the second model representation being configured to produce a second analytics output having a second quality when executed, the first and second model representations being different manifestations of a same model and the second quality being higher than the first quality; receiving, from a user equipment, a request for the second analytics output; and signalling, to the user equipment, at least one of said second analytics output and/or indication of a location of the second model representation.

[0054] The method may comprise generating the first and second model representations from the same model.

[0055]The method may comprise: receiving, from a network function or directly from a user equipment, a request for a user equipment to receive said same model; and signalling metadata identifying at least the first model representation to the network function or the user equipment.

[0056] Said signalling metadata may further comprise signalling metadata identifying a third model representation configured to produce a third analytics output having a third quality when executed, the third model representation being a manifestation of the same model, and the third accuracy having a value between a value for the first accuracy and a value for the second accuracy.

[0057] The method may comprise selecting the first model representation in dependence on at least one of a processing capability of the user equipment and/or a network capacity for downloading the first model representation to the user equipment. [0058]The method may comprise: receiving a request to execute the second model representation; and executing the second model representation using the first analytics output as an input to the second model representation when the request comprises the first analytics output, and/or executing the second model representation using input data that is common to both the first and second model representations. [0059]The request for the second analytics output may comprise said metadata. The request for the second analytics output may comprise the first analytics output.

[0060] The metadata may comprise an address, and wherein said downloading may comprise downloading the first model representation from said address.

[0061]The metadata may comprise at least one of: an indication of how resource intensive at least one of the model representations is to execute relative to other model representations of said model; an indication of an address from which the first model representation may be downloaded; an indication of a version of the first model representation to be executed; an indication of a version of said model; an indication of a type of said model; an indication of an address from which a second model representation may be downloaded, the second model representation being another manifestation of said model and having a second analytics output associated with a second accuracy, the second accuracy being higher than the first accuracy; an indication of an address from which a third model representation may be downloaded, the third model representation being another manifestation of said model and having a third analytics output associated with a third accuracy, the third accuracy being between the first accuracy and the second accuracy; an indication of an address from which a fourth model representation may be downloaded, the fourth model representation being another manifestation of said model and having a fourth analytics output associated with a fourth accuracy, the fourth accuracy being higher than the second accuracy; and metadata related to analytics output(s) of at least one of the first to fourth model representations.

[0062] The network analytics function may comprise at least one of a network data analytics function, a management data analytics service, and/or any network function or network application function configured to perform analytics.

[0063]According to an eighth aspect, there is provided a method for an apparatus for a user equipment, the method comprising: executing a first model representation to produce a first analytics output having a first quality; determining that a second analytics output having a second quality, higher than the first quality, is desired; signalling, to a network analytics function, a request for the second analytics output, from the network analytics function, at least one of said second analytics output and/or indication of a location of a second model representation for obtaining the second analytics output.

[0064]The method may comprise: signalling, to a network function and/or to the network analytics function, a request for a user equipment to receive a model for obtaining an analytics output; receiving, from the network function and/or from the network analytics function, metadata identifying at least the first model representation in response to the signalled request; and downloading the first model representation in response to receiving the metadata.

[0065]The request for the second analytics output may comprise said metadata. The request for the second analytics output may comprise the first analytics output.

[0066] The metadata may comprise an address, and wherein said downloading may comprise downloading the first model representation from said address.

[0067]The method may comprise receiving, from the network function and/or from the network analytics function, metadata identifying a third model representation configured to produce a third analytics output having a third quality when executed, the third model representation being a manifestation of said model, and the third accuracy having a value between a value for the first accuracy and a value for the second accuracy.

[0068]The method may comprise signalling a capability of the user equipment to download and/or execute a model to a network function.

[0069]The capability may be communicated in a registration message.

[0070] The metadata may comprise at least one of: an indication of how resource intensive at least one of the model representations is to execute relative to other model representations of said model; an indication of an address from which the first model representation may be downloaded; an indication of a version of the first model representation to be executed; an indication of a version of said model; an indication of a type of said model; an indication of an address from which a second model representation may be downloaded, the second model representation being another manifestation of said model and having a second analytics output associated with a second accuracy, the second accuracy being higher than the first accuracy; an indication of an address from which a third model representation may be downloaded, the third model representation being another manifestation of said model and having a third analytics output associated with a third accuracy, the third accuracy being between the first accuracy and the second accuracy; an indication of an address from which a fourth model representation may be downloaded, the fourth model representation being another manifestation of said model and having a fourth analytics output associated with a fourth accuracy, the fourth accuracy being higher than the second accuracy; and metadata related to analytics output(s) of at least one of the first to fourth model representations.

[0071]The network analytics function may comprise at least one of a network data analytics function, a management data analytics service, and/or any network function or network application function configured to perform analytics.

[0072]According to a ninth aspect, there is provided a method for an apparatus for a network function, the method comprising: receiving, from a user equipment, a request for the user equipment to receive a model for obtaining an analytics output; signalling the received request to a network analytics function; receiving, from the network analytics function, metadata identifying a first model representation in response to the signalled request, wherein the first model is a manifestation of said model; and signalling the received metadata to the user equipment.

[0073]The method may comprise: receiving, from the user equipment, a capability of the user equipment to download and/or execute said model; and signalling the capability to another network function.

[0074]The capability may be communicated in a registration message.

[0075] The metadata may comprise at least one of: an indication of how resource intensive at least one of the model representations is to execute relative to other model representations of said model; an indication of an address from which the first model representation may be downloaded; an indication of a version of the first model representation to be executed; an indication of a version of said model; an indication of a type of said model; an indication of an address from which a second model representation may be downloaded, the second model representation being another manifestation of said model and having a second analytics output associated with a second accuracy, the second accuracy being higher than the first accuracy; an indication of an address from which a third model representation may be downloaded, the third model representation being another manifestation of said model and having a third analytics output associated with a third accuracy, the third accuracy being between the first accuracy and the second accuracy; an indication of an address from which a fourth model representation may be downloaded, the fourth model representation being another manifestation of said model and having a fourth analytics output associated with a fourth accuracy, the fourth accuracy being higher than the second accuracy; and metadata related to analytics output(s) of at least one of the first to fourth model representations.

[0076] The network analytics function may comprise at least one of a network data analytics function, a management data analytics service, and/or any network function or network application function configured to perform analytics.

[0077]According to a tenth aspect, there is provided an apparatus for a network analytics function, the apparatus comprising: causing circuitry for causing a first model representation to be stored, the first model representation being configured to produce a first analytics output having a first quality when executed; causing circuitry for causing a second model representation to be stored, the second model representation being configured to produce a second analytics output having a second quality when executed, the first and second model representations being different manifestations of a same model and the second quality being higher than the first quality; receiving circuitry for receiving, from a user equipment, a request for the second analytics output; and signalling circuitry for signalling, to the user equipment, at least one of said second analytics output and/or indication of a location of the second model representation.

[0078] The apparatus may comprise generating circuitry for generating the first and second model representations from the same model.

[0079] The apparatus may comprise: receiving circuitry for receiving, from a network function or directly from a user equipment, a request for a user equipment to receive said same model; and signalling circuitry for signalling metadata identifying at least the first model representation to the network function or the user equipment.

[0080] Said signalling circuitry for signalling metadata may further comprise signalling circuitry for signalling metadata identifying a third model representation configured to produce a third analytics output having a third quality when executed, the third model representation being a manifestation of the same model, and the third accuracy having a value between a value for the first accuracy and a value for the second accuracy. [0081]The apparatus may comprise selecting circuitry for selecting the first model representation in dependence on at least one of a processing capability of the user equipment and/or a network capacity for downloading the first model representation to the user equipment.

[0082]The apparatus may comprise: receiving circuitry for receiving a request to execute the second model representation; and executing circuitry for executing the second model representation using the first analytics output as an input to the second model representation when the request comprises the first analytics output, and/or for executing the second model representation using input data that is common to both the first and second model representations.

[0083]The request for the second analytics output may comprise said metadata. The request for the second analytics output may comprise the first analytics output.

[0084] The metadata may comprise an address, and wherein said downloading may comprise downloading the first model representation from said address.

[0085] The metadata may comprise at least one of: an indication of how resource intensive at least one of the model representations is to execute relative to other model representations of said model; an indication of an address from which the first model representation may be downloaded; an indication of a version of the first model representation to be executed; an indication of a version of said model; an indication of a type of said model; an indication of an address from which a second model representation may be downloaded, the second model representation being another manifestation of said model and having a second analytics output associated with a second accuracy, the second accuracy being higher than the first accuracy; an indication of an address from which a third model representation may be downloaded, the third model representation being another manifestation of said model and having a third analytics output associated with a third accuracy, the third accuracy being between the first accuracy and the second accuracy; an indication of an address from which a fourth model representation may be downloaded, the fourth model representation being another manifestation of said model and having a fourth analytics output associated with a fourth accuracy, the fourth accuracy being higher than the second accuracy; and metadata related to analytics output(s) of at least one of the first to fourth model representations.

[0086] The network analytics function may comprise at least one of a network data analytics function, a management data analytics service, and/or any network function or network application function configured to perform analytics. [0087]According to an eleventh aspect, there is provided an apparatus for a user equipment, the apparatus comprising: executing circuitry for executing a first model representation to produce a first analytics output having a first quality; determining circuitry for determining that a second analytics output having a second quality, higher than the first quality, is desired; signalling circuitry for signalling, to a network analytics function, a request for the second analytics output; and receiving circuitry for receiving, from the network analytics function, at least one of said second analytics output and/or indication of a location of a second model representation for obtaining the second analytics output.

[0088] The apparatus may comprise: signalling circuitry for signalling, to a network function and/or to the network analytics function, a request for a user equipment to receive a model for obtaining an analytics output; receiving circuitry for receiving, from the network function and/or from the network analytics function, metadata identifying at least the first model representation in response to the signalled request; and downloading circuitry for downloading the first model representation in response to receiving the metadata.

[0089] The request for the second analytics output may comprise said metadata. The request for the second analytics output may comprise the first analytics output.

[0090] The metadata may comprise an address, and wherein said downloading may comprise downloading the first model representation from said address.

[0091]The apparatus may comprise receiving circuitry for receiving, from the network function and/or from the network analytics function, metadata identifying a third model representation configured to produce a third analytics output having a third quality when executed, the third model representation being a manifestation of said model, and the third accuracy having a value between a value for the first accuracy and a value for the second accuracy.

[0092] The apparatus may comprise signalling circuitry for signalling a capability of the user equipment to download and/or execute a model to a network function.

[0093]The capability may be communicated in a registration message.

[0094] The metadata may comprise at least one of: an indication of how resource intensive at least one of the model representations is to execute relative to other model representations of said model; an indication of an address from which the first model representation may be downloaded; an indication of a version of the first model representation to be executed; an indication of a version of said model; an indication of a type of said model; an indication of an address from which a second model representation may be downloaded, the second model representation being another manifestation of said model and having a second analytics output associated with a second accuracy, the second accuracy being higher than the first accuracy; an indication of an address from which a third model representation may be downloaded, the third model representation being another manifestation of said model and having a third analytics output associated with a third accuracy, the third accuracy being between the first accuracy and the second accuracy; an indication of an address from which a fourth model representation may be downloaded, the fourth model representation being another manifestation of said model and having a fourth analytics output associated with a fourth accuracy, the fourth accuracy being higher than the second accuracy; and metadata related to analytics output(s) of at least one of the first to fourth model representations.

[0095] The network analytics function may comprise at least one of a network data analytics function, a management data analytics service, and/or any network function or network application function configured to perform analytics.

[0096]According to a twelfth aspect, there is provided an apparatus for a network function, the apparatus comprising: receiving circuitry for receiving, from a user equipment, a request for the user equipment to receive a model for obtaining an analytics output; signalling circuitry for signalling the received request to a network analytics function; receiving circuitry for receiving, from the network analytics function, metadata identifying a first model representation in response to the signalled request, wherein the first model is a manifestation of said model; and signalling circuitry for signalling the received metadata to the user equipment.

[0097] The apparatus may comprise: receiving circuitry for receiving, from the user equipment, a capability of the user equipment to download and/or execute said model; and signalling circuitry for signalling the capability to another network function.

[0098]The capability may be communicated in a registration message.

[0099] The metadata may comprise at least one of: an indication of how resource intensive at least one of the model representations is to execute relative to other model representations of said model; an indication of an address from which the first model representation may be downloaded; an indication of a version of the first model representation to be executed; an indication of a version of said model; an indication of a type of said model; an indication of an address from which a second model representation may be downloaded, the second model representation being another manifestation of said model and having a second analytics output associated with a second accuracy, the second accuracy being higher than the first accuracy; an indication of an address from which a third model representation may be downloaded, the third model representation being another manifestation of said model and having a third analytics output associated with a third accuracy, the third accuracy being between the first accuracy and the second accuracy; an indication of an address from which a fourth model representation may be downloaded, the fourth model representation being another manifestation of said model and having a fourth analytics output associated with a fourth accuracy, the fourth accuracy being higher than the second accuracy; and metadata related to analytics output(s) of at least one of the first to fourth model representations.

[0100] The network analytics function may comprise at least one of a network data analytics function, a management data analytics service, and/or any network function or network application function configured to perform analytics.

[0101] According to a thirteenth aspect, there is provided non-transitory computer readable medium comprising program instructions for causing an apparatus for a network analytics function to perform at least the following: cause a first model representation to be stored, the first model representation being configured to produce a first analytics output having a first quality when executed; cause a second model representation to be stored, the second model representation being configured to produce a second analytics output having a second quality when executed, the first and second model representations being different manifestations of a same model and the second quality being higher than the first quality; receive, from a user equipment, a request for the second analytics output; and signal, to the user equipment, at least one of said second analytics output and/or indication of a location of the second model representation.

[0102] The apparatus may be caused to generate the first and second model representations from the same model.

[0103] The apparatus may be caused to: receive, from a network function or directly from a user equipment, a request for a user equipment to receive said same model; and signal metadata identifying at least the first model representation to the network function or the user equipment. [0104] Said signalling metadata may comprise signalling metadata identifying a third model representation configured to produce a third analytics output having a third quality when executed, the third model representation being a manifestation of the same model, and the third accuracy having a value between a value for the first accuracy and a value for the second accuracy.

[0105] The apparatus may be caused to select the first model representation in dependence on at least one of a processing capability of the user equipment and/or a network capacity for downloading the first model representation to the user equipment. [0106] The apparatus may be caused to: receive a request to execute the second model representation; and execute the second model representation using the first analytics output as an input to the second model representation when the request comprises the first analytics output, and/or execute the second model representation using input data that is common to both the first and second model representations. [0107]The request for the second analytics output may comprise said metadata. The request for the second analytics output may comprise the first analytics output.

[0108] The metadata may comprise an address, and wherein said downloading may comprise downloading the first model representation from said address.

[0109] The metadata may comprise at least one of: an indication of how resource intensive at least one of the model representations is to execute relative to other model representations of said model; an indication of an address from which the first model representation may be downloaded; an indication of a version of the first model representation to be executed; an indication of a version of said model; an indication of a type of said model; an indication of an address from which a second model representation may be downloaded, the second model representation being another manifestation of said model and having a second analytics output associated with a second accuracy, the second accuracy being higher than the first accuracy; an indication of an address from which a third model representation may be downloaded, the third model representation being another manifestation of said model and having a third analytics output associated with a third accuracy, the third accuracy being between the first accuracy and the second accuracy; an indication of an address from which a fourth model representation may be downloaded, the fourth model representation being another manifestation of said model and having a fourth analytics output associated with a fourth accuracy, the fourth accuracy being higher than the second accuracy; and metadata related to analytics output(s) of at least one of the first to fourth model representations.

[0110] The network analytics function may comprise at least one of a network data analytics function, a management data analytics service, and/or any network function or network application function configured to perform analytics.

[0111] According to a fourteenth aspect, there is provided non-transitory computer readable medium comprising program instructions for causing an apparatus for a user equipment to perform at least the following: execute a first model representation to produce a first analytics output having a first quality; determine that a second analytics output having a second quality, higher than the first quality, is desired; signal, to a network analytics function, a request for the second analytics output; and receive, from the network analytics function, at least one of said second analytics output and/or indication of a location of a second model representation for obtaining the second analytics output.

[0112] The apparatus may be caused to: signal, to a network function and/or to the network analytics function, a request for a user equipment to receive a model for obtaining an analytics output; receive, from the network function and/or from the network analytics function, metadata identifying at least the first model representation in response to the signalled request; and download the first model representation in response to receiving the metadata.

[0113]The request for the second analytics output may comprise said metadata. The request for the second analytics output may comprise the first analytics output.

[0114]The metadata may comprise an address, and wherein said downloading may comprise downloading the first model representation from said address.

[0115] The apparatus may be caused to receive, from the network function and/or from the network analytics function, metadata identifying a third model representation configured to produce a third analytics output having a third quality when executed, the third model representation being a manifestation of said model, and the third accuracy having a value between a value for the first accuracy and a value for the second accuracy.

[0116] The apparatus may be caused to signal a capability of the user equipment to download and/or execute a model to a network function.

[0117] The capability may be communicated in a registration message. [0118] The metadata may comprise at least one of: an indication of how resource intensive at least one of the model representations is to execute relative to other model representations of said model; an indication of an address from which the first model representation may be downloaded; an indication of a version of the first model representation to be executed; an indication of a version of said model; an indication of a type of said model; an indication of an address from which a second model representation may be downloaded, the second model representation being another manifestation of said model and having a second analytics output associated with a second accuracy, the second accuracy being higher than the first accuracy; an indication of an address from which a third model representation may be downloaded, the third model representation being another manifestation of said model and having a third analytics output associated with a third accuracy, the third accuracy being between the first accuracy and the second accuracy; an indication of an address from which a fourth model representation may be downloaded, the fourth model representation being another manifestation of said model and having a fourth analytics output associated with a fourth accuracy, the fourth accuracy being higher than the second accuracy; and metadata related to analytics output(s) of at least one of the first to fourth model representations.

[0119] The network analytics function may comprise at least one of a network data analytics function, a management data analytics service, and/or any network function or network application function configured to perform analytics.

[0120] According to a fifteenth aspect, there is provided non-transitory computer readable medium comprising program instructions for causing an apparatus for a network function to perform at least the following: receive, from a user equipment, a request for the user equipment to receive a model for obtaining an analytics output; signal the received request to a network analytics function; receive, from the network analytics function, metadata identifying a first model representation in response to the signalled request, wherein the first model is a manifestation of said model; and signal the received metadata to the user equipment.

[0121]The apparatus may be caused to: receive, from the user equipment, a capability of the user equipment to download and/or execute said model; and signal the capability to another network function.

[0122]The capability may be communicated in a registration message. [0123] The metadata may comprise at least one of: an indication of how resource intensive at least one of the model representations is to execute relative to other model representations of said model; an indication of an address from which the first model representation may be downloaded; an indication of a version of the first model representation to be executed; an indication of a version of said model; an indication of a type of said model; an indication of an address from which a second model representation may be downloaded, the second model representation being another manifestation of said model and having a second analytics output associated with a second accuracy, the second accuracy being higher than the first accuracy; an indication of an address from which a third model representation may be downloaded, the third model representation being another manifestation of said model and having a third analytics output associated with a third accuracy, the third accuracy being between the first accuracy and the second accuracy; an indication of an address from which a fourth model representation may be downloaded, the fourth model representation being another manifestation of said model and having a fourth analytics output associated with a fourth accuracy, the fourth accuracy being higher than the second accuracy; and metadata related to analytics output(s) of at least one of the first to fourth model representations.

[0124] The network analytics function may comprise at least one of a network data analytics function, a management data analytics service, and/or any network function or network application function configured to perform analytics.

[0125] According to a sixteenth aspect, there is provided a computer program product stored on a medium that may cause an apparatus to perform any method as described herein.

[0126] According to a seventeenth aspect, there is provided an electronic device that may comprise apparatus as described herein.

[0127] According to an eighteenth aspect, there is provided a chipset that may comprise an apparatus as described herein.

Brief description of Figures

[0128] Examples will now be described, by way of example only, with reference to the accompanying Figures in which:

[0129] Figures 1A and 1 B show a schematic representation of a 5G system;

[0130] Figure 2 shows a schematic representation of a network apparatus; [0131] Figure 3 shows a schematic representation of a user equipment;

[0132] Figure 4 shows a schematic representation of a non-volatile memory medium storing instructions which when executed by a processor allow a processor to perform one or more of the steps of the methods of some examples;

[0133] Figure 5 shows a schematic representation of a network;

[0134] Figures 6A and 6B illustrate different machine learning and/or artificial intelligence learning algorithms;

[0135] Figure 7 is an example signalling mechanism;

[0136] Figure 8 shows a schematic representation of a configuration of apparatus; and [0137] Figures 9 to 11 are flow charts illustrating operations that may be performed by apparatus described herein.

Detailed description

[0138] In the following, certain aspects are explained with reference to mobile communication devices capable of communication via a wireless cellular system and mobile communication systems serving such communication devices. For brevity and clarity, the following describes such aspects with reference to a 5G wireless communication system. However, it is understood that such aspects are not limited to 5G wireless communication systems, and may, for example, be applied to other wireless communication systems with analogous components (for example, current 6G proposals).

[0139] Before explaining in detail the exemplifying embodiments, certain general principles of a 5G wireless communication system are briefly explained with reference to Figures 1A and 1 B.

[0140] Figure 1A shows a schematic representation of a 5G system (5GS) 100. The 5GS may comprise a user equipment (UE) 102 (which may also be referred to as a communication device or a terminal), a 5G access network (AN) (which may be a 5G Radio Access Network (RAN) or any other type of 5G AN such as a Non-3GPP Interworking Function (N3IWF) /a Trusted Non3GPP Gateway Function (TNGF) for Untrusted / Trusted Non-3GPP access or Wireline Access Gateway Function (W-AGF) for Wireline access) 104, a 5G core (5GC) 106, one or more application functions (AF) 108 and one or more data networks (DN) 110. [0141]The 5G RAN may comprise one or more gNodeB (gNB) distributed unit functions connected to one or more gNodeB (gNB) unit functions. The RAN may comprise one or more access nodes.

[0142]The 5GC 106 may comprise one or more Access and Mobility Management Functions (AMF) 112, one or more Session Management Functions (SMF) 114, one or more authentication server functions (ALISF) 116, one or more unified data management (UDM) functions 118, one or more user plane functions (UPF) 120, one or more unified data repository (UDR) functions 122, one or more network repository functions (NRF) 128, and/or one or more network exposure functions (NEF) 124. Although NRF 128 is not depicted with its interfaces, it is understood that this is for clarity reasons and that NRF 128 may have a plurality of interfaces with other network functions.

[0143] The 5GC 106 also comprises a network data analytics function (NWDAF) 126. The NWDAF is responsible for providing network analytics information upon request from one or more network functions or apparatus within the network. Network functions can also subscribe to the NWDAF 126 to receive information therefrom. Accordingly, the NWDAF 126 is also configured to receive and store network information from one or more network functions or apparatus within the network. The data collection by the NWDAF 126 may be performed based on at least one subscription to the events provided by the at least one network function.

[0144] The network may further comprise a management data analytics service (MDAS). The MDAS may provide data analytics of different network related parameters including for example load level and/or resource utilisation. For example, the MDAS for a network function (NF) can collect the NF’s load related performance data, e.g., resource usage status of the NF. The analysis of the collected data may provide forecast of resource usage information in a predefined future time. This analysis may also recommend appropriate actions e.g., scaling of resources, admission control, load balancing of traffic, etc.

[0145] Figure 1 B shows a schematic representation of a 5GC 106’ represented in current 3GPP specifications.

[0146] Figure 1 B shows a UPF 120’ connected to an SMF 114’ over an N4 interface. The SMF 114’ is connected to each of a UDR 122’, an NEF 124’, an NWDAF 126’, an AF 108’, a Policy Control Function (PCF) 130’, an AMF 112’, and a Charging function 132’ over an interconnect medium that also connects these network functions to each other.

[0147]3GPP refers to a group of organizations that develop and release different standardized communication protocols. 3GPP is currently developing and publishing documents related to Release 16, relating to 5G technology, with Release 17 currently being scheduled for 2022.

[0148] Current 3GPP networks utilize machine learning (ML) and/or artificial intelligence (Al) models for performing analytics within the network. The results of such analytics (i.e. the outputs of the model(s), also referred to herein as analytics output) may be used for making decisions in the network, such as with respect to resource usage, connection/handover control, etc. The output of a model may be statistical (e.g. providing an historical overview of network behavior). The output of a model may be an inference (e.g. providing a prediction for a future network behavior based on current information and previous trends). It is understood that any reference below to a particular type of analytics output (e.g. prediction, inference and/or statistical) includes references to other types of analytics output.

[0149] Some disclosures have considered how a UE can obtain analytics. As one example disclosure, a UE can invoke any analytics of network data at an appropriate network function (e.g., the NWDAF), which produces an analytics output for the requesting UE. As one example, the NWDAF and/or the MDAS prepares a model, based on AI/ML, that outputs a generated output. The requesting UE may then consume the analytics output. In these existing disclosures, the analytics is performed wholly at the network, and without any assistance from the UE.

[0150] Preparing models for such a purpose can take a considerable amount of time, especially when there is a large number of variables in the model. For example, consider the case of a model based on a Network Function (NF) and/or an application function (AF) being interested in predicting a UE's outgoing call pattern (i.e., how many calls a UE is predicted to make in the next hour). In this case, the longer the model preparation phase (i.e. the more training data is considered), the more accurate the ultimate model prediction is. It is therefore desirable to take longer on collecting data and preparing the model when aiming to obtain an accurate output/prediction.

[0151] Moreover, a big model (example size 200 MB+) cannot be shared efficiently over Radio Access Network (RAN) signalling as it takes up too many transmission resources. For example, service-based Interface (SBI) signalling can accommodate only a maximum of 16 MB of data per message. SBI is the term given to the Application Program Interface (API)-based communication that can take place between two Virtualized Network Functions (VNFs) within the 5G Service Based Architecture. A given VNF can utilise an API call over the SBI in order to invoke a particular service or service operation.

[0152] UEs, e.g. robots or device UEs, may have a powerful processing power, and may perform analytics by themselves. For example, a camera (i.e. a type of UE) installed in the police department may be configured to process an image/video by itself (i.e. without network-based processing), and only output a result when a suspect is detected in the image/video who matches a record in the police's database.

[0153] In NWDAF/MDAS functions, the 3GPP network has a very powerful analytics engine that can host and train AI/ML models. However, the analytics power of the NWDAF/MDAS will be limited by a number of different factors. For example, the analytics power may be limited by the hardware resources available to the NWDAF/MDAS. As another example, the analytics power may be limited by the quantity of data being collected from different UEs, especially when larger quantities of data are being collected. As another example, the analytics power may be limited by at least one of the network and/or UE bandwidth for data transfer. There are lots of UEs connected to the network, which may be able to provide additional resources to support the analytics generation.

[0154] At the same time, if model training at the NWDAF/MDAS requires input data from the UEs, a huge amount of data might be collected from the UEs, whereby partial training executed at the UE can reduce the amount of data to be transferred from the UE to the network. The 3GPP network does not currently utilize concepts of split/distributed training of Figures 6A to 6B to leverage the UE capability to train the model and expand the resources for model training. These concepts are illustrated further below.

[0155] Figure 6A illustrates a plurality of UEs 601 a configured to communicate, via an access point 602a, with a network 603a. In this case, AI/ML learning operation is split between the different endpoints.

[0156] In the system of Figure 6A, the AI/ML operations are split between different AI/ML endpoints (e.g. device, base station, data center server). This means that UEs and network elements cooperate with each other to efficiently perform training and/or inference. This involves a tradeoff between amount of computation at edge devices and data communication at radio links.

[0157] Figure 6B illustrates a plurality of UEs 601 b that are configured to receive a model from a network 603b via an access point 602b. In this case, the full AI/ML model data is distributed to each UE 601 b.

[0158] This example of Figure 6B is also referred to as model distribution over 5GS, and involves sharing a trained model from a model provider entity (e.g., in the network) to one or more analytics entities (e.g. UEs I UE applications). Currently, 3GPP (as of Rel.17) only specifies ML model sharing between NWDAF instances, which means that each instance of the NWDAF may employ the received model on its own to locally produce analytics outputs.

[0159] A lightweight model may be employed by the UE for a first analysis of the input data. A lightweight model may be considered to be a model having an output associated with a lower accuracy, where the lower accuracy is relative to the outputs of heavier models for the inference. For example, if the output is an image, the image output for a lightweight model may have a grainier appearance (i.e. a larger pixel size) than the output from a heavier weight model would. When a higher accuracy is required after the model is executed at the UE, the complete input data could be sent to a network server to run a heavier (i.e. a more accurate) model. However, this would lead to a huge amount of input data being sent to the network server. Further, the network server runs the heavier model based on the available input data, without being able to leverage some intermediate results available from the UE based on the inference with the lightweight model. In another example, if the output is an object recognition, the output for a lightweight model may be limited to detecting the rough shape of an object (e.g. rectangular), while a heavier weight model can identify specific objects (e.g. a chair). In another example, the output for a lightweight model has a higher error rate compared to the output from a heavier weight model.

[0160] UE resource limitation is a challenge: It will not always be feasible for a network to send a full/heavier model to a UE to handle. It would therefore be useful for some lightweight version of the model to be processed at the UE, with the heavier version of the model being processed at the network if needed.

[0161]The following discloses a framework in which a network can create both a lightweight AI/ML model and more sophisticated/heavier models to perform a same inference task. This is referred to herein as "progressive inference", and the presently disclosed framework allows both a UE and a network server to progressively (e.g. step-wise) make use of both computational and communication resources to get inference results based on a desired inference quality requirement.

[0162] In particular, the lightweight model can be provided to a UE while the other, heavier models are stored in the network. The UE may then run the lightweight model locally. If inference results are not up to the desired quality, the UE may invoke other models available in the network to generate a more accurate output based on the inference result created by the UE. This can be done by, for example, the UE providing a set of requirements to the network, such as a minimum accuracy of output to be obtained from the heavier weight model. As another example, this may be done by the UE sequentially asking the network for inference results with a better quality. The network may provide authorization of model invocation, perform inference, and provide the more accurate inference results to the UE.

[0163] This technique is further illustrated with respect to Figures 7 and 8.

[0164] Figure 7 is a signalling diagram that shows signalling between a UE 701 , an AMF 702, an SMF or User Plane Function (UPF) 703, an NWDAF 704, an Application Server (AS) 705, a UDM 706 and a UDR 707.

[0165] At 7001 , the NWDAF 704 prepares a model for obtaining an output inference.

[0166]To illustrate how this may be done, the following considers the example of an image processing application with respect to Figure 8. However, it is understood that principles described below may be applied to other types of AI/ML models, particularly where inference output results of different quality may be used at different times.

[0167] Figure 8 illustrates a UE 801 , a Radio access network 802 and a core network 803. The core network comprises at least one of an NWDAF 804 and an MDAS 804 (shown as one entity in Figure 9 to illustrate that either entity, or both, may perform the presently described functions). The NWDAF/MDAS 804 comprises three versions of the same model, each version being associated with a respective accuracy.

[0168] In a more detailed example, the core network may be configured to provide an image processing model that can detect an image (for example, the above-mentioned face detections at the police station). However, the full version of this model (also referred to herein as the full model) is not shared with the UE 801 , 701 as result of its large size, the limited UE capacity, and/or because this model is only to be used for a single, one-off inference task. As the full version is not being shared, the NWDAF/MDAS 804 divides the model into three separate model representations, which are also referred to herein as manifestations and/or layers. The manifestations may also be thought of as instantiations of the full model. Each of the separate model representations is configured to produce a respective output having an associated accuracy that is different to the accuracy output by the full model. It is understood that the number of layers the model is divided into may be other than three. It is understood that the number of layers/representations the model is divided into may be variable, and/or selectable by the NWDAF/MDAS in dependence on the model itself. These model layers are also referred to herein as model representations, with a model representation being a manifestation of the full model with which it is associated. These model manifestations/representations may also be referred to and/or labelled as being an expression of the full model, and/or a realization of the full model, and/or an implementation of the full mode. Each model layer can perform inference to some degree, and respectively generate an output result. These model layers will be labelled herein as Model Representation 1 , Model Representation 2, and Model Representation 3 to distinguish between them.

[0169] Model Representation 1 is a lightweight model that is lighter (e.g. in size and/or required computational resources) than each of Model Representation 2 and Model Representation 3. In other words, Models 2 and 3 are considered to be heavier models than Model Representation 1 , as they use more computational resources to output an inference, and/or because they are larger in size. How lightweight Model Representation 1 is selected and/or generated may be determined considering the limited resources available, e.g. UE processing power, UE storage capacity, network capacity, etc.

[0170]The NWDAF/MDAS 804 may push at least Model Representation 1 (i.e. the lightest model) to an application server from which it can be downloaded by the UEs. [0171] Model Representation 1 may comprise meta data that links Model Representation 1 to other models (e.g. Model Representation 2 and Model Representation 3, implementing additional Model Representations 2 and 3). This meta data may comprise at least one of: o An indication that it is the lightest weight model, which is referred to herein as Model Representation (MR1 ) o An address (or some other unique identifier) for Model Representation 1 o Version of model (i.e. a version number associated with the full model manifested by Model Representation 1 ) o Model Representation version o Type of model (e.g. that the model is considered to be “lightweight” or “heavy weight”, or the analytics type the model is based on) o An address (or some other unique identifier) of Model Representation 2 (MR2)/Model Representation 2. The address may be, for example, a Uniform Resource Indicator (URI) o An address (or some other unique identifier) of Model Representation 3 (MR3)/Model Representation 3. The address may be, for example, a Uniform Resource Indicator (URI) o Output parameters related metadata (e.g. what kind of parameters are obtained at the output of MR1 ). These may be, for example, at least one of:

■ A number of input samples processed to form the output

■ A time taken to process those input samples

■ A generated output satisfaction ratio indicating how satisfactory the output of MR1 is.

[0172] Steps 7002 to 7007 relate to establishing a UE capability of the UE 701 and storing this information in the network. In particular, at least one network element in the core network may store a UE capability (example: AMF, UDM/UDR). This stored capability may later be used by the network when it wants to push the model to the compatible UEs.

[0173] At 7002 the UE 701 signals to the AMF 702 whether the UE is capable of downloading a model for performing analytics. This capability may be signalled in, for example, a registration request message. When the UE is capable of downlink such a model, the UE may additionally comprise an indication of its available resources for executing the model. The indication of the available resources may be, for example, an indication of processing resources available at the UE 701 and/or an indication of a downlink transmission rate available for transmission of the model. When the UE is signalling that it is capable of processing the analytics and/or more on its own, the capability may be signalled using UEModel/AnalyticsProcessingCapability = True. When the UE is signalling that it is not capable of processing the analytics and/or more on its own, the capability may be signalled using UEModel/AnalyticsProcessingCapability = False. In other words, the indication may have selectable values, one value indicating that the UE is capable, and an alternate value indicating the UE is incapable. It is understood that although the presently described mechanism illustrates an example in which the UE capability is pushed to the AMF, that the capability may instead be pulled to a network function in response to a query to the UE from the network function.

[0174] At 7003, the AMF 702 signals the UDM 706. This signalling of 7003 may comprise an indication of the UE capability provided in 7002. When the signalling of 7002 is a registration message, this signalling may be an Nudm_UECM_Registration message. The signalling of 7003 may comprise a subscriber identifier for the UE 701 . For example, the signalling of 7003 may comprise 5G globally unique Subscription Permanent Identifier (SUPI). The signalling of 7003 may comprise an identifier of the AMF 702. For example, the signalling of 7003 may comprise a Globally Unique AMF ID (GUAMI).

[0175] At 7004, the UDM 706 signals the UDR 707. This signalling of 7004 may comprise the indication of the user capability received in 7003. This signalling may be sent as part of, for example, an Nudr update procedure.

[0176] At 7005, the UDM 706 responds to the signalling of 7003. This response may acknowledge receipt of the signalling of 7003.

[0177] At 7006, the UDM 706 and the AMF 702 may exchange signalling. For example, where the signalling of 7002 from the UE 701 was a registration request and/or a subscription request, the signalling may relate to registering the UE and/or subscribing the UE 701 to requested events. Where the request is a subscription request, the signalling exchanged may be, for example, an Nudm_SDM_Subscribe message, such as is currently defined in 3GPP TS 23.502.

[0178] At 7007, the AMF responds to the signalling of 7002. This response may comprise an indication that the UE is allowed to use analytics. This may be indicated by setting a value in an information element (called herein “UsingNetworkAnalyticsAllowed”) in the response to indicate true. When the UE is not allowed to use analytics, this information element may instead have its value set to indicate false. Whether true or false is indicated in this information element may be set in dependence on a determination of a network element. This determination may take into consideration the UE 701 location and/or the network operator’s policy regarding UEs performing analytical work. When 7002 was a registration request message, the signalling of 7007 may be, for example, a registration accept message.

[0179] Steps 7008 to 7013 relate to the UE 701 downloading a model to be executed by the UE 701. [0180] At 7008, the UE 701 determines that it has, or would like a subscription for a processing model. In the present example, this processing model is a model for identifying an object in an image.

[0181] At 7009, the UE 701 signals the AMF 702. This signalling is a request for the model associated with the determined description. This signalling may be NAS signalling. The NAS signalling may be provided using a new information element to what has previously been known.

[0182] At 7010, the AMF 702 signals the NWDAF 704. This signalling comprises the model request of 7009. This signalling may be performed after the AMF 702 has successfully authorised the request received in 7009. This authorisation may be based, for example, on subscription data for the requesting UE 701 . This authorisation may alternatively or additionally be based on operator policy for execution of a model. The NWDAF 704 selects a model (Model Representation 1 ) for the UE 701 in response to receipt of the signalling of 7010. The selection may be performed by taking into account the capabilities of the UE 701 (e.g. with respect to processing power). The selection may additionally be performed by taking into account a target accuracy for the output of the model that is desired by the UE 701 .

[0183]At 7011 , the NWDAF 704 returns meta data associated with Model Representation 1 to the AMF 702. In other words, the NWDAF 704 returns meta data for the lightest model to the AMF 704 in response to the model request of 7010. The metadata may identify the model to which the UE has a subscription. The metadata may identify a version of the model to which the UE has a subscription. For example, the metadata may identify Model Representation 1 to the UE 701.

[0184] At 7012, the AMF 702 forwards the received meta data of 7011 to the UE 701. [0185] At 7013, the UE 701 downloads the model associated by the meta data of 7011. To this effect, the received meta data may comprise an address from which the associated model may be downloaded by the UE 701 .

[0186] Steps 7014 to 7017 relate to the execution of the model/obtaining results from the model.

[0187] At 7014, the UE feeds the input (e.g. an image in the present case) into the model downloaded at 7013 to obtain an output (also referred to herein as an inference). The UE 701 further determines if the obtained output fulfils a quality threshold. This may be performed, for example, by determining an estimated accuracy associated to a threshold accuracy, with the threshold accuracy being predetermined according to the purpose for the model being executed. The accuracy of the model may be determined after the model is run.

[0188] For example, if a police department UE is examining an image to detect a face based on Model Representation 1 available at the UE, then Model Representation 1 might generate some basic face recognition, while it cannot generate the detailed/precise/complete output (i.e. a clear face detection based on hundreds of facial features). When the UE cannot generate an output of a high enough quality, the UE may invoke the execution of at least one of Model Representation 2 and/or Model Representation 3 at the network. How the UE determines whether the estimated accuracy fulfils the quality threshold may be implementation and use case specific. For example, in this face recognition example, only when the MR1 identifies a face in the video stream will the UE 701 send this data to the network for further recognition using a heavier weight model (e.g. male vs female, age, ... ).

[0189] Therefore, at 7015, the UE 701 signals the Model Representation 1 output obtained by the UE 701 to the NWDAF 704. This signalling may further indicate that a better quality output than the obtained output is requested by the UE 701. The signalling may further comprise metadata associated with Model Representation 1 . For example, the signalling may comprise at least some of the metadata received at 7012.

[0190]The signalling of 7015 may identify a specific version of the model to be used for producing an output. For example, the UE may identify Model Representation 2 in the request of 7015. As another example, the UE may identify Model Representation 3 in the request of 7015. It is understood that multiple models (e.g. Model Representation 2 and Model Representation 3) may be identified in the signalling of 7015.

[0191]The model(s) requested may be identified by, for example, a unique identifier, such as an address of that model. The address may be, for example, a URI. The model(s) to be requested may be selected by the UE 701 in dependence on the result of a determination of the UE with respect to how much more accurate the output needs to be. As an alternative to identifying specific models or in combination thereof, the UE may signal an indication of a quality of output to be obtained for the UE’s particular application of the model. In this case, the NWDAF may use the signalled indication of quality to select a Model (e.g. Model Representation 2 and/or Model Representation 3) to be executed for meeting that signalled indication of quality. [0192] The signalling of 7015 may be routed via the user plane. The signalling of 7015 may be routed via the control plane. When the signalling is routed via the control plane, then the control plane may be used for signalling metadata information while the user plane may be used for signalling a model or for performing a big data transfer.

[0193] At 7016, the NWDAF 702 uses the received Model Representation 1 output as an input into at least one of Model Representation 2 and/or Model Representation 3. The NWDAF may, using the received Model Representation 1 output as an input to Model Representation 2, obtain a Model Representation 2 output. When the Model Representation 2 output satisfies a quality requested in the signalling of 7015, the Model Representation 2 output may be returned to the UE 701 in 7017. When the Model Representation 2 output does not satisfy the quality requested in the signalling of 7015, the Model Representation 2 output may be used as an input to Model Representation 3, which produces a Model Representation 3 output. In this case, the Model Representation 3 output may be returned to the UE 701 at 7017. It is understood that both the Model Representation 2 output and the Model Representation 3 output may be provided to the UE 701 in 7017.

[0194]There are numerous advantages to the presently described system. For example, it enables a UE without a network coverage to perform inference by itself based on lightweight Model Representations/models downloaded to the UE in advance.

[0195] Further, when the network is heavily congested, local inference by the UE using a lightweight model may help reduce traffic load on radio links and improve the overall Quality of Service for the network.

[0196] Figures 9 to 11 are flow charts illustrating potential operations that may be performed by the apparatus described herein. These apparatus may be configured to interact with each other. It is understood that the following highlights certain features that are described in the example above, and that other features from the above examples may be implemented in the presently described systems.

[0197] Figure 9 illustrates operations that may be performed by an apparatus for a network analytics function. The network analytics function may be at least one of a NWDAF, a MDAS, and/or any network function or network application function configured to perform analytics. [0198] At 901 , the apparatus causes a first model representation to be stored, the first model representation being configured to produce a first analytics output having a first quality when executed.

[0199] At 902, the apparatus causes a second model representation to be stored, the second model representation being configured to produce a second analytics output having a second quality when executed. The second quality is higher than the first quality.

[0200] The first and second model representations are different manifestations of a same model. By this, it is understood that the first and second model representations may be configured to provide different approximations directed towards determining at least part of an overarching objective of the same model. For example, where the aim is to identify a particular face in an image, the first model may simply be configured to identify a face, while the second model may be configured to identify properties of the identified face. Therefore, each manifestation may represent a respective set of criteria for achieving an objective of the same model.

[0201] At 903, the apparatus receives, from a user equipment, a request for the second analytics output.

[0202]At 904, the apparatus signals, to the user equipment, at least one of said second analytics output and/or indication of a location of the second model representation. This signalling may be performed in response to receipt of the request of 903.

[0203] Whether the second analytics output and/or the indication of the location is sent in 904 may be determined in dependence on a determination of the user equipment’s capability for executing the second model representation within, such as a determination of the user equipment’s processing resources.

[0204] The apparatus may be configured to generate at least the first and second model representations from the same model. Where third and/or fourth model representations are available, these may also be generated from the same model. The number of model representations may be higher than four.

[0205] The apparatus may be configured to receive, from a network function, a request for a user equipment to receive said same model. In response to receipt of this request, the apparatus may signal metadata identifying at least the first model representation to the network function and/or to the user equipment. [0206] The apparatus may be configured to receive, directly from a user equipment, a request for the user equipment to receive said same model. In response to receipt of this request, the apparatus may signal metadata identifying at least the first model representation to the user equipment.

[0207] In both/either of these cases, signalling metadata may further comprise signalling metadata identifying a third model representation configured to produce a third analytics output having a third quality when executed, the third model representation being a manifestation of the same model, and the third accuracy having a value between a value for the first accuracy and a value for the second accuracy. This signalling of the third model representation information may be performed at a different time to the signalling of the first model representation. Alternatively, the signalling of the third model representation information may be performed at the same time as the signalling of the first model representation.

[0208] The apparatus may select the first model representation (and/or the third model representation, where applicable) in dependence on at least one of a processing capability of the user equipment and/or a network capacity for downloading the first model representation (and/or the third model representation, where applicable) to the user equipment.

[0209]The request for the second analytics output may comprise said metadata. The request for the second analytics output may comprise the first analytics output.

[0210] The metadata may comprise at least one of: an indication of how resource intensive the first model representation is to execute relative to other model representations of the same model; an indication of an address from which the first model representation may be downloaded; an indication of a version of the first model representation to be executed; an indication of a version of the same model; an indication of a type of the same model; an indication of an address from which the second model representation may be downloaded; an indication of an address from which the third model representation may be downloaded; an indication of an address from which a fourth model representation may be downloaded, the fourth model representation being another manifestation of the same model and having a fourth analytics output associated with a fourth accuracy, the fourth accuracy being higher than the second accuracy; and metadata related to analytics output(s) of at least one of the first to fourth model representations. [0211]The apparatus may be configured to, in response to receiving a request to execute the second model representation, execute the second model representation using the first analytics output as an input to the second model representation when the request comprises the first analytics output. Alternatively or in addition, the apparatus may execute the second model representation using input data that is common to both the first and second model representations.

[0212] Figure 10 illustrates potential operations that may be performed by, for example, an apparatus for a user equipment. The user equipment may be the user equipment mentioned above in relation to Figure 9.

[0213] At 1001 , the apparatus executes a first model representation to produce a first analytics output having a first quality.

[0214] At 1002, the apparatus may determine that a second analytics output having a second quality, higher than the first quality, is desired. This may be determined by, for example, comparing the analytics output to a defined objective that the user equipment is configured to achieve. This defined objective may be expressed by a set of criteria configured in the user equipment.

[0215] At 1003, the apparatus signals, to a network analytics function, a request for the second analytics output. The request may be transmitted to the network analytics function directly or indirectly (e.g. though the network function of Figure 11 ).

[0216] At 1004, the apparatus receives, from the network analytics function, at least one of said second analytics output and/or indication of a location of a second model representation for obtaining the second analytics output.

[0217] The apparatus may signal, to a network function and/or to the network analytics function, a request for a user equipment to receive a model for obtaining an analytics output. In response to this request, the apparatus may receive, from the network function and/or from the network analytics function, metadata identifying at least the first model representation in response to the signalled request. The apparatus may download the first model representation in response to receiving the metadata.

[0218]The request for the second analytics output may comprise said metadata. The request for the second analytics output may comprise the first analytics output.

[0219] The metadata may comprise an address, and said downloading may comprise downloading the first model representation from said address.

[0220]The apparatus may receive, from the network function and/or from the network analytics function, metadata identifying a third model representation configured to produce a third analytics output having a third quality when executed. The third model representation may be a manifestation of said model, and the third accuracy having a value between a value for the first accuracy and a value for the second accuracy.

[0221]The metadata may comprise at least one of: an indication of how resource intensive the first model representation is to execute relative to other model representations of said model; an indication of an address from which the first model representation may be downloaded; an indication of a version of the first model representation to be executed; an indication of a version of said model; an indication of a type of said model; an indication of an address from which the second model representation may be downloaded; an indication of an address from which the third model representation may be downloaded; an indication of an address from which a fourth model representation may be downloaded, the fourth model representation being another manifestation of said model and having a fourth analytics output associated with a fourth accuracy, the fourth accuracy being higher than the second accuracy; and metadata related to analytics output(s) of at least one of the first to fourth model representations.

[0222] The apparatus may signal a capability of the user equipment to download and/or execute a model to a network function. The signalling the capability may comprise signalling the capability in a registration message.

[0223] Figure 11 illustrates operations that may be performed by an apparatus for a network function. The network function may be the network function referred to above in relation to Figures 9 and 10.

[0224] At 1101 , the apparatus receives, from a user equipment, a request for the user equipment to receive a model for obtaining an analytics output. The user equipment may be the user equipment of Figure 10.

[0225] At 1102, the apparatus signals the received request to a network analytics function. The network analytics function may be the network analytics function of Figure 9. This signalling of 1102 may be performed responsive to the signalling of 1101.

[0226] At 1103, the user equipment receives, from the network analytics function, metadata identifying a first model representation in response to the signalled request, wherein the first model is a manifestation of said model.

[0227] At 1104, the network apparatus signals the received metadata to the user equipment. [0228]The apparatus may receive, from the user equipment, a capability of the user equipment to download and/or execute said model. The capability may be as described above in relation to Figures 9 and 10. The apparatus may signal the capability to another network function, such as to the network analytics function.

[0229] The apparatus may receive the capability in a registration message. The registration message may be a registration message signalled by the user equipment when the user equipment first requests to use the network. This registration message may be signalled using non-access stratum signalling.

[0230] The metadata may comprise at least one of: an indication of how resource intensive the first model representation is to execute relative to other model representations of said model; an indication of an address from which the first model representation may be downloaded; an indication of a version of the first model representation to be executed; an indication of a version of said model; an indication of a type of said model; an indication of an address from which a second model representation may be downloaded, the second model representation being another manifestation of said model and having a second analytics output associated with a second accuracy, the second accuracy being higher than the first accuracy; an indication of an address from which the third model representation may be downloaded; an indication of an address from which a fourth model representation may be downloaded, the fourth model representation being another manifestation of said model and having a fourth analytics output associated with a fourth accuracy, the fourth accuracy being higher than the second accuracy; and metadata related to analytics output(s) of at least one of the first to fourth model representations.

[0231] Figure 2 shows an example of a control apparatus for a communication system, for example to be coupled to and/or for controlling a station of an access system, such as a RAN node, e.g. a base station, gNB, a central unit of a cloud architecture or a node of a core network such as an MME or S-GW, a scheduling entity such as a spectrum management entity, or a server or host, for example an apparatus hosting an NRF, NWDAF, AMF, SMF, UDM/UDR etc. The control apparatus may be integrated with or external to a node or module of a core network or RAN. In some embodiments, base stations comprise a separate control apparatus unit or module. In other embodiments, the control apparatus can be another network element such as a radio network controller or a spectrum controller. The control apparatus 200 can be arranged to provide control on communications in the service area of the system. The apparatus 200 comprises at least one memory 201 , at least one data processing unit 202, 203 and an input/output interface 204. Via the interface the control apparatus can be coupled to a receiver and a transmitter of the apparatus. The receiver and/or the transmitter may be implemented as a radio front end or a remote radio head. For example the control apparatus 200 or processor 201 can be configured to execute an appropriate software code to provide the control functions.

[0232]A possible wireless communication device will now be described in more detail with reference to Figure 3 showing a schematic, partially sectioned view of a communication device 300. Such a communication device is often referred to as user equipment (UE) or terminal. An appropriate mobile communication device may be provided by any device capable of sending and receiving radio signals. Non-limiting examples comprise a mobile station (MS) or mobile device such as a mobile phone or what is known as a ’smart phone’, a computer provided with a wireless interface card or other wireless interface facility (e.g., USB dongle), personal data assistant (PDA) or a tablet provided with wireless communication capabilities, or any combinations of these or the like. A mobile communication device may provide, for example, communication of data for carrying communications such as voice, electronic mail (email), text message, multimedia and so on. Users may thus be offered and provided numerous services via their communication devices. Non-limiting examples of these services comprise two-way or multi-way calls, data communication or multimedia services or simply an access to a data communications network system, such as the Internet. Users may also be provided broadcast or multicast data. Non-limiting examples of the content comprise downloads, television and radio programs, videos, advertisements, various alerts and other information.

[0233]A wireless communication device may be for example a mobile device, that is, a device not fixed to a particular location, or it may be a stationary device. The wireless device may need human interaction for communication, or may not need human interaction for communication. In the present teachings the terms UE or “user” are used to refer to any type of wireless communication device.

[0234] The wireless device 300 may receive signals over an air or radio interface 307 via appropriate apparatus for receiving and may transmit signals via appropriate apparatus for transmitting radio signals. In Figure 3 transceiver apparatus is designated schematically by block 306. The transceiver apparatus 306 may be provided for example by means of a radio part and associated antenna arrangement. The antenna arrangement may be arranged internally or externally to the wireless device.

[0235]A wireless device is typically provided with at least one data processing entity 301 , at least one memory 302 and other possible components 303 for use in software and hardware aided execution of tasks it is designed to perform, including control of access to and communications with access systems and other communication devices. The data processing, storage and other relevant control apparatus can be provided on an appropriate circuit board and/or in chipsets. This feature is denoted by reference 704. The user may control the operation of the wireless device by means of a suitable user interface such as key pad 305, voice commands, touch sensitive screen or pad, combinations thereof or the like. A display 308, a speaker and a microphone can be also provided. Furthermore, a wireless communication device may comprise appropriate connectors (either wired or wireless) to other devices and/or for connecting external accessories, for example hands-free equipment, thereto.

[0236] Figure 4 shows a schematic representation of non-volatile memory media 400a (e.g. computer disc (CD) or digital versatile disc (DVD)) and 400b (e.g. universal serial bus (USB) memory stick) storing instructions and/or parameters 402 which when executed by a processor allow the processor to perform one or more of the steps of the methods of Figure 9 and/or Figure 10, and/or Figure 11 .

[0237] The embodiments may thus vary within the scope of the attached claims. In general, some embodiments may be implemented in hardware or special purpose circuits, software, logic or any combination thereof. For example, some aspects may be implemented in hardware, while other aspects may be implemented in firmware or software which may be executed by a controller, microprocessor or other computing device, although embodiments are not limited thereto. While various embodiments may be illustrated and described as block diagrams, flow charts, or using some other pictorial representation, it is well understood that these blocks, apparatus, systems, techniques or methods described herein may be implemented in, as non-limiting examples, hardware, software, firmware, special purpose circuits or logic, general purpose hardware or controller or other computing devices, or some combination thereof.

[0238]The embodiments may be implemented by computer software stored in a memory and executable by at least one data processor of the involved entities or by hardware, or by a combination of software and hardware. Further in this regard it should be noted that any procedures, e.g., as in Figure 9 and/or Figure 10, and/or Figure 11 , may represent program steps, or interconnected logic circuits, blocks and functions, or a combination of program steps and logic circuits, blocks and functions. The software may be stored on such physical media as memory chips, or memory blocks implemented within the processor, magnetic media such as hard disk or floppy disks, and optical media such as for example DVD and the data variants thereof, CD. [0239] The memory may be of any type suitable to the local technical environment and may be implemented using any suitable data storage technology, such as semiconductor-based memory devices, magnetic memory devices and systems, optical memory devices and systems, fixed memory and removable memory. The data processors may be of any type suitable to the local technical environment, and may include one or more of general purpose computers, special purpose computers, microprocessors, digital signal processors (DSPs), application specific integrated circuits (AStudy ItemC), gate level circuits and processors based on multi-core processor architecture, as non-limiting examples.

[0240] Alternatively or additionally some embodiments may be implemented using circuitry. The circuitry may be configured to perform one or more of the functions and/or method steps previously described. That circuitry may be provided in the base station and/or in the communications device.

[0241]As used in this application, the term “circuitry” may refer to one or more or all of the following:

(a) hardware-only circuit implementations (such as implementations in only analogue and/or digital circuitry);

(b) combinations of hardware circuits and software, such as:

(i) a combination of analogue and/or digital hardware circuit(s) with software/firmware and

(ii) any portions of hardware processor(s) with software (including digital signal processor(s)), software, and memory(ies) that work together to cause an apparatus, such as the communications device or base station to perform the various functions previously described; and

(c) hardware circuit(s) and or processor(s), such as a microprocessor(s) or a portion of a microprocessor(s), that requires software (e.g., firmware) for operation, but the software may not be present when it is not needed for operation. [0242] This definition of circuitry applies to all uses of this term in this application, including in any claims. As a further example, as used in this application, the term circuitry also covers an implementation of merely a hardware circuit or processor (or multiple processors) or portion of a hardware circuit or processor and its (or their) accompanying software and/or firmware. The term circuitry also covers, for example integrated device.

[0243]The foregoing description has provided by way of exemplary and non-limiting examples a full and informative description of some embodiments. However, various modifications and adaptations may become apparent to those skilled in the relevant arts in view of the foregoing description, when read in conjunction with the accompanying drawings and the appended claims. However, all such and similar modifications of the teachings will still fall within the scope as defined in the appended claims.

[0244] In the above, different examples are described using, as an example of an access architecture to which the presently described techniques may be applied, a radio access architecture based on long term evolution advanced (LTE Advanced, LTE-A) or new radio (NR, 5G), without restricting the examples to such an architecture, however. The examples may also be applied to other kinds of communications networks having suitable means by adjusting parameters and procedures appropriately. Some examples of other options for suitable systems are the universal mobile telecommunications system (UMTS) radio access network (UTRAN), wireless local area network (WLAN or WiFi), worldwide interoperability for microwave access (WiMAX), Bluetooth®, personal communications services (PCS), ZigBee®, wideband code division multiple access (WCDMA), systems using ultra-wideband (UWB) technology, sensor networks, mobile ad-hoc networks (MANETs) and Internet Protocol multimedia subsystems (IMS) or any combination thereof.

[0245] Figure 5 depicts examples of simplified system architectures only showing some elements and functional entities, all being logical units, whose implementation may differ from what is shown. The connections shown in Figure 5 are logical connections; the actual physical connections may be different. It is apparent to a person skilled in the art that the system typically comprises also other functions and structures than those shown in Figure 5. [0246]The examples are not, however, restricted to the system given as an example but a person skilled in the art may apply the solution to other communication systems provided with necessary properties.

[0247] The example of Figure 5 shows a part of an exemplifying radio access network. For example, the radio access network may support sidelink communications described below in more detail.

[0248] Figure 5 shows devices 500 and 502. The devices 500 and 502 are configured to be in a wireless connection on one or more communication channels with a node 504. The node 504 is further connected to a core network 506. In one example, the node 504 may be an access node such as (eZg)NodeB serving devices in a cell. In one example, the node 504 may be a non-3GPP access node. The physical link from a device to a (eZg)NodeB is called uplink or reverse link and the physical link from the (eZg)NodeB to the device is called downlink or forward link. It should be appreciated that (eZg)NodeBs or their functionalities may be implemented by using any node, host, server or access point etc. entity suitable for such a usage.

[0249]A communications system typically comprises more than one (eZg)NodeB in which case the (eZg)NodeBs may also be configured to communicate with one another over links, wired or wireless, designed for the purpose. These links may be used for signalling purposes. The (eZg)NodeB is a computing device configured to control the radio resources of communication system it is coupled to. The NodeB may also be referred to as a base station, an access point or any other type of interfacing device including a relay station capable of operating in a wireless environment. The (eZg)NodeB includes or is coupled to transceivers. From the transceivers of the (eZg)NodeB, a connection is provided to an antenna unit that establishes bi-directional radio links to devices. The antenna unit may comprise a plurality of antennas or antenna elements. The (eZg)NodeB is further connected to the core network 506 (CN or next generation core NGC). Depending on the deployed technology, the (eZg)NodeB is connected to a serving and packet data network gateway (S-GW +P-GW) or user plane function (UPF), for routing and forwarding user data packets and for providing connectivity of devices to one or more external packet data networks, and to a mobile management entity (MME) or access mobility management function (AMF), for controlling access and mobility of the devices.

[0250] Examples of a device are a subscriber unit, a user device, a user equipment (UE), a user terminal, a terminal device, a mobile station, a mobile device, etc [0251]The device typically refers to a mobile or static device ( e.g. a portable or nonportable computing device) that includes wireless mobile communication devices operating with or without an universal subscriber identification module (IISIM), including, but not limited to, the following types of devices: mobile phone, smartphone, personal digital assistant (PDA), handset, device using a wireless modem (alarm or measurement device, etc.), laptop and/or touch screen computer, tablet, game console, notebook, and multimedia device. It should be appreciated that a device may also be a nearly exclusive uplink only device, of which an example is a camera or video camera loading images or video clips to a network. A device may also be a device having capability to operate in Internet of Things (loT) network which is a scenario in which objects are provided with the ability to transfer data over a network without requiring human-to-human or human-to-computer interaction, e.g. to be used in smart power grids and connected vehicles. The device may also utilise cloud. In some applications, a device may comprise a user portable device with radio parts (such as a watch, earphones or eyeglasses) and the computation is carried out in the cloud.

[0252] The device illustrates one type of an apparatus to which resources on the air interface are allocated and assigned, and thus any feature described herein with a device may be implemented with a corresponding apparatus, such as a relay node. An example of such a relay node is a layer 3 relay (self-backhauling relay) towards the base station. The device (or, in some examples, a layer 3 relay node) is configured to perform one or more of user equipment functionalities.

[0253]Various techniques described herein may also be applied to a cyber-physical system (CPS) (a system of collaborating computational elements controlling physical entities). CPS may enable the implementation and exploitation of massive amounts of interconnected information and communications technology, ICT, devices (sensors, actuators, processors microcontrollers, etc.) embedded in physical objects at different locations. Mobile cyber physical systems, in which the physical system in question has inherent mobility, are a subcategory of cyber-physical systems. Examples of mobile physical systems include mobile robotics and electronics transported by humans or animals.

[0254] Additionally, although the apparatuses have been depicted as single entities, different units, processors and/or memory units (not all shown in Figure 5) may be implemented. [0255] 5G enables using multiple input - multiple output (MIMO) antennas, many more base stations or nodes than the LTE (a so-called small cell concept), including macro sites operating in co-operation with smaller stations and employing a variety of radio technologies depending on service needs, use cases and/or spectrum available. 5G mobile communications supports a wide range of use cases and related applications including video streaming, augmented reality, different ways of data sharing and various forms of machine type applications (such as (massive) machine-type communications (mMTC), including vehicular safety, different sensors and real-time control). 5G is expected to have multiple radio interfaces, e.g. below 6GHz or above 24 GHz, cm Wave and mmWave, and also being integrable with existing legacy radio access technologies, such as the LTE. Integration with the LTE may be implemented, at least in the early phase, as a system, where macro coverage is provided by the LTE and 5G radio interface access comes from small cells by aggregation to the LTE. In other words, 5G is planned to support both inter-RAT operability (such as LTE-5G) and inter-RI operability (inter-radio interface operability, such as below 6GHz - cmWave, 6 or above 24 GHz - cmWave and mmWave). One of the concepts considered to be used in 5G networks is network slicing in which multiple independent and dedicated virtual sub-networks (network instances) may be created within the same infrastructure to run services that have different requirements on latency, reliability, throughput and mobility.

[0256] The current architecture in LTE networks is fully distributed in the radio and fully centralized in the core network. The low latency applications and services in 5G require to bring the content close to the radio which leads to local break out and multiaccess edge computing (MEC). 5G enables analytics and knowledge generation to occur at the source of the data. This approach requires leveraging resources that may not be continuously connected to a network such as laptops, smartphones, tablets and sensors. MEC provides a distributed computing environment for application and service hosting. It also has the ability to store and process content in close proximity to cellular subscribers for faster response time. Edge computing covers a wide range of technologies such as wireless sensor networks, mobile data acquisition, mobile signature analysis, cooperative distributed peer-to-peer ad hoc networking and processing also classifiable as local cloud/fog computing and grid/mesh computing, dew computing, mobile edge computing, cloudlet, distributed data storage and retrieval, autonomic self-healing networks, remote cloud services, augmented and virtual reality, data caching, Internet of Things (massive connectivity and/or latency critical), critical communications (autonomous vehicles, traffic safety, real-time analytics, time-critical control, healthcare applications).

[0257]The communication system is also able to communicate with other networks 512, such as a public switched telephone network, or a VoIP network, or the Internet, or a private network, or utilize services provided by them. The communication network may also be able to support the usage of cloud services, for example at least part of core network operations may be carried out as a cloud service (this is depicted in Figure 5 by “cloud” 514). The communication system may also comprise a central control entity, or a like, providing facilities for networks of different operators to cooperate for example in spectrum sharing.

[0258] The technology of Edge cloud may be brought into a radio access network (RAN) by utilizing network function virtualization (NFV) and software defined networking (SDN). Using the technology of edge cloud may mean access node operations to be carried out, at least partly, in a server, host or node operationally coupled to a remote radio head or base station comprising radio parts. It is also possible that node operations will be distributed among a plurality of servers, nodes or hosts. Application of cloudRAN architecture enables RAN real time functions being carried out at or close to a remote antenna site (in a distributed unit, DU 508) and non- real time functions being carried out in a centralized manner (in a centralized unit, CU 510).

[0259] It should also be understood that the distribution of labour between core network operations and base station operations may differ from that of the LTE or even be non-existent. Some other technology advancements probably to be used are Big Data and all-IP, which may change the way networks are being constructed and managed. 5G (or new radio, NR) networks are being designed to support multiple hierarchies, where MEC servers can be placed between the core and the base station or nodeB (gNB). It should be appreciated that MEC can be applied in 4G networks as well.

[0260] 5G may also utilize satellite communication to enhance or complement the coverage of 5G service, for example by providing backhauling. Possible use cases are providing service continuity for machine-to-machine (M2M) or Internet of Things (loT) devices or for passengers on board of vehicles, Mobile Broadband, (MBB) or ensuring service availability for critical communications, and future railway/maritime/aeronautical communications. Satellite communication may utilise geostationary earth orbit (GEO) satellite systems, but also low earth orbit (LEO) satellite systems, in particular mega-constellations (systems in which hundreds of (nano)satellites are deployed). Each satellite in the mega-constellation may cover several satellite-enabled network entities that create on-ground cells. The on-ground cells may be created through an on-ground relay node or by a gNB located on-ground or in a satellite.

[0261] It is obvious for a person skilled in the art that the depicted system is only an example of a part of a radio access system and in practice, the system may comprise a plurality of (eZg)NodeBs, the device may have an access to a plurality of radio cells and the system may comprise also other apparatuses, such as physical layer relay nodes or other network elements, etc. At least one of the (eZg)NodeBs or may be a Home(eZg)nodeB. Additionally, in a geographical area of a radio communication system a plurality of different kinds of radio cells as well as a plurality of radio cells may be provided. Radio cells may be macro cells (or umbrella cells) which are large cells, usually having a diameter of up to tens of kilometers, or smaller cells such as micro-, femto- or picocells. The (eZg)NodeBs of Figure 5 may provide any kind of these cells. A cellular radio system may be implemented as a multilayer network including several kinds of cells. Typically, in multilayer networks, one access node provides one kind of a cell or cells, and thus a plurality of (eZg)NodeBs are required to provide such a network structure.

[0262] For fulfilling the need for improving the deployment and performance of communication systems, the concept of “plug-and-play” (eZg)NodeBs has been introduced. Typically, a network which is able to use “plug-and-play” (eZg)Node Bs, includes, in addition to Home (eZg)NodeBs (H(eZg)nodeBs), a home node B gateway, or HNB-GW (not shown in Figure 5). A HNB Gateway (HNB-GW), which is typically installed within an operator’s network may aggregate traffic from a large number of HNBs back to a core network.

Claims

1 ) An apparatus for a network analytics function, the apparatus comprising means for: causing a first model representation to be stored, the first model representation being configured to produce a first analytics output having a first quality when executed; causing a second model representation to be stored, the second model representation being configured to produce a second analytics output having a second quality when executed, the first and second model representations being different manifestations of a same model and the second quality being higher than the first quality; receiving, from a user equipment, a request for the second analytics output; and signalling, to the user equipment, at least one of said second analytics output and/or indication of a location of the second model representation.

2) An apparatus as claimed in claim 1 , comprising means for generating the first and second model representations from the same model.

3) An apparatus as claimed in any preceding claim, comprising means for: receiving, from a network function or directly from a user equipment, a request for a user equipment to receive said same model; and signalling metadata identifying at least the first model representation to the network function or the user equipment.

4) An apparatus as claimed in claim 3, wherein said means for signalling metadata further comprises means for signalling metadata identifying a third model representation configured to produce a third analytics output having a third quality when executed, the third model representation being a manifestation of the same model, and the third accuracy having a value between a value for the first accuracy and a value for the second accuracy.

49 ) An apparatus as claimed in claim 3 or 4, comprising means for selecting the first model representation in dependence on at least one of a processing capability of the user equipment and/or a network capacity for downloading the first model representation to the user equipment. ) An apparatus as claimed in any preceding claim, comprising means for: receiving a request to execute the second model representation; and executing the second model representation using the first analytics output as an input to the second model representation when the request comprises the first analytics output, and/or executing the second model representation using input data that is common to both the first and second model representations. ) An apparatus for a user equipment, the apparatus comprising means for: executing a first model representation to produce a first analytics output having a first quality; determining that a second analytics output having a second quality, higher than the first quality, is desired; signalling, to a network analytics function, a request for the second analytics output; and receiving, from the network analytics function, at least one of said second analytics output and/or indication of a location of a second model representation for obtaining the second analytics output. ) An apparatus as claimed in claim 7, comprising means for: signalling, to a network function and/or to the network analytics function, a request for a user equipment to receive a model for obtaining an analytics output; receiving, from the network function and/or from the network analytics function, metadata identifying at least the first model representation in response to the signalled request; and downloading the first model representation in response to receiving the metadata.

50 ) An apparatus as claimed in claim 3 or 8, or any claim dependent on claim 3 or 8, wherein the request for the second analytics output comprises said metadata and/or the request for the second analytics output comprises the first analytics output,. 0)An apparatus as claimed in any of claims 7 to 9, wherein the metadata comprises an address, and wherein said downloading comprises downloading the first model representation from said address. 1 )An apparatus as claimed in any of claims 7 to 10, comprising means for receiving, from the network function and/or from the network analytics function, metadata identifying a third model representation configured to produce a third analytics output having a third quality when executed, the third model representation being a manifestation of said model, and the third accuracy having a value between a value for the first accuracy and a value for the second accuracy. 2)An apparatus as claimed in any of claims 7 to 11 , comprising means for signalling a capability of the user equipment to download and/or execute a model to a network function. 3)An apparatus for a network function, the apparatus comprising means for: receiving, from a user equipment, a request for the user equipment to receive a model for obtaining an analytics output; signalling the received request to a network analytics function; receiving, from the network analytics function, metadata identifying a first model representation in response to the signalled request, wherein the first model is a manifestation of said model; and signalling the received metadata to the user equipment. 4)An apparatus as claimed in claim 13, comprising means for: receiving, from the user equipment, a capability of the user equipment to download and/or execute said model; and signalling the capability to another network function.

51 )An apparatus as claimed in claim 12 or 14, wherein the capability is communicated in a registration message. )An apparatus as claimed in any of claims 3, 8 or 14, or any claim dependent on claims 3, 8 or 14, wherein the metadata comprises at least one of: an indication of how resource intensive at least one of the model representations is to execute relative to other model representations of said model; an indication of an address from which the first model representation may be downloaded; an indication of a version of the first model representation to be executed; an indication of a version of said model; an indication of a type of said model; an indication of an address from which a second model representation may be downloaded, the second model representation being another manifestation of said model and having a second analytics output associated with a second accuracy, the second accuracy being higher than the first accuracy; an indication of an address from which a third model representation may be downloaded, the third model representation being another manifestation of said model and having a third analytics output associated with a third accuracy, the third accuracy being between the first accuracy and the second accuracy; an indication of an address from which a fourth model representation may be downloaded, the fourth model representation being another manifestation of said model and having a fourth analytics output associated with a fourth accuracy, the fourth accuracy being higher than the second accuracy; and metadata related to analytics output(s) of at least one of the first to fourth model representations. )An apparatus as claimed in any preceding claim, wherein the network analytics function comprises at least one of a network data analytics function, a management data analytics service, and/or any network function or network application function configured to perform analytics. ) A method for network analytics, the method comprising:

52 storing a first model representation, the first model representation being configured to produce a first analytics output having a first quality when executed; storing a second model representation to be stored, the second model representation being configured to produce a second analytics output having a second quality when executed, the first and second model representations being different manifestations of a same model and the second quality being higher than the first quality; receiving, from a user equipment, a request for the second analytics output, the request comprising the first analytics output; and signalling, to the user equipment, at least one of said second analytics output and/or indication of a location of the second model representation. ) A method in a user equipment, the method comprising: executing a first model representation to produce a first analytics output having a first quality; determining that a second analytics output having a second quality, higher than the first quality, is desired; signalling, to a network analytics function, a request for the second analytics output, the request comprising the first analytics output; and receiving, from the network analytics function, at least one of said second analytics output and/or indication of a location of a second model representation for obtaining the second analytics output. ) A method in a network function, the method comprising: receiving, from a user equipment, a request for the user equipment to receive a model for obtaining an analytics output; signalling the received request to a network analytics function; receiving, from the network analytics function, metadata identifying a first model representation in response to the signalled request, wherein the first model is a manifestation of said model; and signalling the received metadata to the user equipment.