Method, apparatus and computer program

Abstract

An apparatus (e.g. source) provides to a data consumer an attribute(s) indicating, for each of a plurality of data formats, whether the apparatus is able to support providing data in said data format.

Description

TECHNICAL FIELD

[0001] Various example embodiments of this disclosure relate to a method, apparatus, system and computer program and in particular but not exclusively to providing a data consumer with an attribute indicating whether the apparatus is able to support data in each of a plurality of different data formats. BACKGROUND [0002JA communication network can be seen as a facility that enables communications between two or more communication devices, or provides communication devices access to a data network. A mobile or wireless communication network is one example of a communication network. A communication device may be provided with a service by an application server.

[0003] Such communication networks operate in according with standards such as those provided by 3GPP (Third Generation Partnership Project) or ETSI (European Telecommunications Standards Institute). Examples of standards are the so-called 5G (5th Generation) standards and 6G (6th Generation) standards provided by 3GPP. SUMMARY

[0004] Some example embodiments of this disclosure will be described with respect to certain aspects. These aspects are not intended to indicate key or essential features of the embodiments of this disclosure, nor are they intended to be used to limit the scope of thereof. Other features, aspects, and elements will be readily apparent to a person skilled in the art in view of this disclosure.

[0005] According to a first aspect, there is provided an apparatus comprising means for: providing, to a data consumer, an attribute indicating, for each of a plurality of data formats, whether the apparatus is able to support providing data in said data format, wherein the plurality of different data formats comprises: raw data, a representation of the raw data, a generative model for regenerating the raw data as synthetic data, and the synthetic data; obtaining, from the data consumer, a request for data to be provided to the data consumer using at least one of said plurality of data formats, the at least one of said plurality of data formats comprising a first data format; and providing the requested data to the data consumer using the first data format.

[0006] According to a second aspect, there is provided an 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 perform: providing, to a data consumer, an attribute indicating, for each of a plurality of data formats, whether the apparatus is able to support providing data in said data format, wherein the plurality of different data formats comprises: raw data, a representation of the raw data, a generative model for regenerating the raw data as synthetic data, and the synthetic data; obtaining, from the data consumer, a request for data to be provided to the data consumer using at least one of said plurality of data formats, the at least one of said plurality of data formats comprising a first data format; and providing the requested data to the data consumer using the first data format.

[0007] According to a third aspect, there is provided a method for an apparatus, the method comprising: providing, to a data consumer, an attribute indicating, for each of a plurality of data formats, whether the apparatus is able to support providing data in said data format, wherein the plurality of different data formats comprises: raw data, a representation of the raw data, a generative model for regenerating the raw data as synthetic data, and the synthetic data; obtaining, from the data consumer, a request for data to be provided to the data consumer using at least one of said plurality of data formats, the at least one of said plurality of data formats comprising a first data format; and providing the requested data to the data consumer using the first data format.

[0008] According to a fourth aspect, there is provided an apparatus comprising: providing circuitry for providing, to a data consumer, an attribute indicating, for each of a plurality of data formats, whether the apparatus is able to support providing data in said data format, wherein the plurality of different data formats comprises: raw data, a representation of the raw data, a generative model for regenerating the raw data as synthetic data, and the synthetic data; obtaining circuitry for obtaining, from the data consumer, a request for data to be provided to the data consumer using at least one of said plurality of data formats, the at least one of said plurality of data formats comprising a first data format; and providing circuitry for providing the requested data to the data consumer using the first data format.

[0009] The following may apply in respect of any (e.g., one or more, including all) of the above first to fourth aspects.

[0010] The apparatus may be caused to perform: selecting the first data format from the at least one of said data formats.

[0011] The apparatus may be caused to perform: evaluating a current and / or estimated property of a communication channel over which the requested data is to be provided, wherein the selecting the first data format is based on the current and / or estimated property and / or wherein the attribute is configured based on the current and / or estimated property, wherein the property comprises at least one of a signal quality, a channel capacity, a channel throughput, or an available bandwidth.

[0012] The apparatus may be caused to perform: receiving an indication of the current and / or estimated property from the data consumer.

[0013] The apparatus may be caused to perform: estimating computing requirements for performing two or more of: storing the raw data, storing the representation of raw data, generating the representation of raw data using an encoder, generating the synthetic data, storing a decoder of the representation, and / or storing the generative model, wherein the selecting the first data format is based on the estimated computing requirements and / or wherein the attribute is configured based on the estimated computing requirements.

[0014] The selecting may further comprise: providing, to the data consumer, an indication of the estimated computing requirements for two or more data formats of said plurality of data formats; obtaining, from the data consumer, an indication of a preferred data format for said providing the requested data; and selecting the first data format based on the indicated preferred data format.

[0015] The selecting may further comprising: obtaining, from the data consumer, an indication of a computing capacity of the data consumer, and wherein the selecting may use the indicated computing capacity and the estimated computing requirements to select first data format, and / or to provide, to the data consumer, an indication of the estimated computing requirements for two or more of said plurality of data formats.

[0016] The apparatus may be caused to perform: tracking an accuracy and / or fidelity of at least one of the generative model or a representation model that forms the representation of raw data over time, wherein the selecting comprises selecting the first data format based on the tracked accuracy and / or wherein the attribute is configured based on the tracked accuracy.

[0017] The tracking the accuracy and / or fidelity may comprise detecting whether the generative model and / or the representation model produces an output that fulfils a first predetermined criteria of accuracy and / or a first predetermined criteria of fidelity, and the apparatus may further be caused to perform, when it is detected that the generative model and / or the representation model does not fulfil the first predetermined criteria of accuracy and / or the first predetermined criteria of fidelity, modifying the generative model and / or the representation model until the first predetermined criteria of accuracy and / or the first predetermined criteria of fidelity is fulfilled.

[0018] The apparatus may be caused to perform: when it is determined that the generative model and / or the representation model fulfils a second predetermined criteria of accuracy and / or a second predetermined criteria of fidelity, removing older data from the generative model and / or the representation model.

[0019] The apparatus may be caused to perform: collecting the raw data; and training at least one of the generative model and / or a representation model that outputs the representation data using the collected raw data.

[0020] The apparatus may be caused to perform, when the first format is the representation of raw data: determining whether to provide a decoder of the raw data to the data consumer, and providing the decoder to the data consumer when it is determined to provide the decoder.

[0021] According to a fifth aspect, there is provided an apparatus comprising means for: obtaining, from a data producer, an attribute indicating, for each of a plurality of data formats, whether the data producer is able to support providing data in said data format, wherein the plurality of different data formats comprises: raw data, a representation of the raw data, a generative model for regenerating the raw data as synthetic data, and the synthetic data; providing, to the data producer, a request for data to be provided to the apparatus using at least one of said plurality of data formats, the at least one of said plurality of data formats comprising a first data format; and obtaining the requested data using the first data format.

[0022] According to a sixth aspect, there is provided an 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 perform: obtaining, from a data producer, an attribute indicating, for each of a plurality of data formats, whether the data producer is able to support providing data in said data format, wherein the plurality of different data formats comprises: raw data, a representation of the raw data, a generative model for regenerating the raw data as synthetic data, and the synthetic data; providing, to the data producer, a request for data to be provided to the apparatus using at least one of said plurality of data formats, the at least one of said plurality of data formats comprising a first data format; and obtaining the requested data using the first data format.

[0023] According to a seventh aspect, there is provided a method for an apparatus, the method comprising: obtaining, from a data producer, an attribute indicating, for each of a plurality of data formats, whether the data producer is able to support providing data in said data format, wherein the plurality of different data formats comprises: raw data, a representation of the raw data, a generative model for regenerating the raw data as synthetic data, and the synthetic data; providing, to the data producer, a request for data to be provided to the apparatus using at least one of said plurality of data formats, the at least one of said plurality of data formats comprising a first data format; and obtaining the requested data using the first data format.

[0024] According to an eighth aspect, there is provided an apparatus comprising: obtaining circuitry for obtaining, from a data producer, an attribute indicating, for each of a plurality of data formats, whether the data producer is able to support providing data in said data format, wherein the plurality of different data formats comprises: raw data, a representation of the raw data, a generative model for regenerating the raw data as synthetic data, and the synthetic data; providing circuitry for providing, to the data producer, a request for data to be provided to the apparatus using at least one of said plurality of data formats, the at least one of said plurality of data formats comprising a first data format; and obtaining circuitry for obtaining the requested data using the first data format.

[0025] The following may apply in respect of any (e.g., one or more, including all) of the above fifth to eighth aspects.

[0026] The apparatus may be caused to perform: before said obtaining the data: obtaining, from the data producer, an indication of estimated computing requirements for two or more of said forms for obtaining said data; selected a preferred form for obtaining said data based on the estimated computing requirements and a current and / or estimated future computing capacity of the apparatus; and providing an indication of the preferred form to the data producer.

[0027] The apparatus may be caused to perform: providing, to the data producer, an indication of a computing capacity of the apparatus.

[0028] The apparatus may be caused to perform: when the first format is the representation of raw data: obtaining a decoder for decoding the representation data from the data producer.

[0029] The following may apply in respect of any (e.g., one or more, including all) of the above first to eighth aspects.

[0030] The requested data may be accompanied by an indication that identifies the first data format.

[0031] The attribute may be provided with an indication that indicates an expected performance of the generative model.

[0032] The request for data may comprise at least one policy that defines at least one constraint and / or condition for selecting the first data format from the plurality of data formats.

[0033] The request may comprise at least one constraint for selecting the first data format from the plurality of data formats, the at least one constraint comprising at least one of: a data type, time granularity, time duration, throughput requirements for data transfer, delay requirements for data transfer, or a request for a specific form for receiving the data.

[0034] The raw data may be the data requested in the request for data.

[0035] The representation of the raw data may comprise an encoded form of the raw data that, when decoded, indicates characteristics of the raw data and / or a reconstructed form of the raw data.

[0036] The generative model may be a model that, when executed, outputs the synthetic data, wherein the synthetic data approximates the raw data within a predetermined accuracy threshold.

[0037] The synthetic data may approximate the raw data within a predetermined accuracy threshold.

[0038] According to an aspect, there is provided a non-transitory computer readable medium comprising program instructions that, when executed by an apparatus, cause the apparatus to perform at least the method according to any of the preceding aspects.

[0039] In the above, many different embodiments have been described. It should be appreciated that further embodiments may be provided by the combination of any two or more of the embodiments described above. DESCRIPTION OF FIGURES

[0040] Some example embodiments will now be described, by way of non-limiting and illustrative example only, with reference to the accompanying Figures in which:

[0041] Figures 1A and 1B shows a representation of a communication system;

[0042] Figure 2 shows a representation of an apparatus for the communication system of Figures 1A and 1B according to some example embodiments;

[0043] Figure 3 shows a representation of an apparatus according to some example embodiments;

[0044] Figures 4A to 4F illustrate example data producer and data consumer configurations;

[0045] Figure 5 illustrates an example communication environment;

[0046] Figures 6 and 7 illustrate example signalling; and

[0047] Figures 8 and 9 illustrate example methods. DETAILED DESCRIPTION

[0048] In general, the following relates to a data producer providing a data consumer with information for making a choice regarding which data format will be used for providing data to that data consumer.

[0049] In more detail, a data producer provides a data consumer with an indication of the different data formats that the data producer is configured to provide data to the data consumer. The different data formats comprise: raw data, a representation of the raw data, a generative model, and synthetic data. These different data formats will be illustrated further below in a more detailed format.

[0050] Different parameters and / or criteria may be used to select which of these four different data formats will be used to provide the requested data to the data consumer. These different parameters may comprise, for example, at least one of: • Data transfer requirements: This parameter relates to a specification for receiving the requested data, and may comprise at factors such as, for example, at least one: the types of data (e.g., traffic class to be received by the data consumer and / or use case for that requested data), time granularity, time duration for providing the requested data, a total amount of data being requested, and data transfer constraints on throughput and delay. • Communication capacity: This communication capacity parameter may consider, for example, network conditions for data transfer over a channel between a data producer and a data consumer, such as bandwidth and delay. Communication capacity parameters are used to estimate whether the requested amount of data can be transferred within the specified data transfer delay constraints. • Computational requirements: Computational resource requirements for storing and executing each information format, including the Central Processing Unit (CPU), Graphics Processing Unit (GPU), hard disk storage, and temporary memory such as Random Access Memory (RAM). • Computational capacity: Computational capacity may comprise, for example, at least one of: CPU resources available at the data consumer and / or at the data producer, GPU resources available at the data consumer and / or at the data producer, hard disk storage available at the data consumer and / or at the data producer, or memory resources available at the data consumer and / or the data producer. • Model quality: Model quality may relate to, for example, an evaluated quality and fidelity of a representation mode (RM) and generative model (GM), and their generated representations and synthetic data, respectively.

[0051] Although these concepts will be illustrated more fully below, the following provides an example of a communication environment in which the presently described techniques may be deployed. It is understood that this communication environment is not limiting, and is merely being used to provide at least one example for describing where such techniques may be deployed.

[0052] Stated differently, in the following various example embodiments are explained with reference to communication devices capable of communication with a communication system. Before explaining in detail the embodiments of the methods and apparatuses of the present disclosure, a 5th generation communication system (5GS), an access network and a core network (5GC) thereof, and communication devices are briefly explained with reference to Figures 1A, 1B, 2 and 3.

[0053] Figure 1A shows a schematic representation of a 5G communication system (5GS). The 5GS may comprise a user equipment (UE), an access network such as a 5G radio access network (5G-RAN) or next generation radio access network (NG-RAN), a 5G core network (5GC), and one or more application functions. An application function may be deployed in the 5GS as trusted application function or may be deployed or host on one or more application servers of the data network. Such application functions are untrusted application functions. The 5GS connects the UE to a data network the access network and the 5GC (e.g., a UPF of the 5GC).

[0054] The 5G-RAN may comprise one or more radio access nodes, such as gNodeB (gNB). A gNB may include one or more gNodeB distributed units connected to one or more gNodeB centralized units. The 5G-RAN may be as illustrated below in Figure 1B.

[0055] The 5GC may comprise the following network functions: Network Slice Selection Function (NSSF); Network Exposure Function; Network Repository Function (NRF); Policy Control Function (PCF); Unified Data Management (UDM); Application Function (AF); Authentication Server Function (AUSF); an Access and Mobility Management Function (AMF); and Session Management Function (SMF), and a user plane function (UPF). Figure 1A also shows the various interfaces (N1, N2 etc.) that may be implemented between the various elements of the system.

[0056] The term “terminal device” refers to any end device that may be capable of wireless communication. By way of example rather than limitation, a terminal device may also be referred to as a communication device, user equipment (UE), a Subscriber Station (SS), a Portable Subscriber Station, a mobile device, a Mobile Station (MS), or an Access Terminal (AT). The terminal device may include, but not limited to, a mobile phone, a cellular phone, a smart phone, voice over IP (VoIP) phones, wireless local loop phones, a tablet, a wearable terminal device, a personal digital assistant (PDA), portable computers, desktop computer, image capture terminal devices such as digital cameras, gaming terminal devices, music storage and playback appliances, vehicle-mounted wireless terminal devices, wireless endpoints, mobile stations, laptop-embedded equipment (LEE), laptop-mounted equipment (LME), USB dongles, smart devices, wireless customer-premises equipment (CPE), a machine-type communications (MTC) device, an Internet of Things (loT) device, a watch or other wearable, a head-mounted display (HMD), a vehicle, a drone, a medical device and applications (e.g., remote surgery), an industrial device and applications (e.g., a robot and / or other wireless devices operating in an industrial and / or an automated processing chain contexts), a data consumer electronics device, a device operating on commercial and / or industrial wireless networks, and the like. The terminal device may also correspond to a Mobile Termination (MT) part of an Integrated Access and Backhaul (IAB) node (e.g., a relay node). In the following description, the terms “terminal device”, “communication device”, “terminal”, “user device”, “user equipment” and “UE” may be used interchangeably.

[0057] As used herein, the term “network device" is used interchangeably with “network access node”, and refers to a node in a communication network via which a terminal device accesses the network and receives services therefrom. The network device may refer to a base station (BS) or an access point (AP), for example, a node B (NodeB or NB), an evolved NodeB (eNodeB or eNB), an NR NB (also referred to as a gNB), a Remote Radio Unit (RRU), a radio header (RH), a remote radio head (RRH), a relay, an Integrated Access and Backhaul node, a low power node such as a femto, a pico, a non-terrestrial network (NTN) or non-ground network device such as a satellite network device, a low earth orbit (LEO) satellite and a geosynchronous earth orbit (GEO) satellite, an aircraft network device, and so forth, depending on the applied terminology and technology. In some example embodiments, radio access network (RAN) split architecture comprises a Centralized Unit (CU) and a Distributed Unit (DU) at an IAB donor node. An IAB node comprises a Mobile Terminal (IAB-MT) part that behaves like a UE toward the parent node, and a DU part of an IAB node behaves like a base station toward the next-hop IAB node.

[0058] In some example embodiments, a link from the network device 120 to the user device 110 or 115 is referred to as a DL, while a link from the user device 110 or 115 to the network device 120 is referred to as a UL. Links are also referred to herein as “channels”. In DL, the network device 120 is a Tx device (or a transmitter), and the user device 110 or 115 is a Rx device (or a receiver). In UL, the user device 110 or 115 is a Tx device (or a transmitter), and the network device 120 is a Rx device (or a receiver). A link between the user device 110 and another user device (not shown) is referred to as a sidelink (SL). In SL, one of the user devices is a Tx device (or a transmitter), and the other of the user devices is a Rx device (or a receiver).

[0059] Communications in the communication environment 100 may be implemented according to any proper communication protocol(s), comprising, but not limited to, cellular communication protocols of the first generation (1G), the second generation (2G), the third generation (3G), the fourth generation (4G), the fifth generation (5G), the sixth generation (6G), and the like, wireless local network communication protocols such as Institute for Electrical and Electronics Engineers (IEEE) 802.11 and the like, and / or any other protocols currently known or to be developed in the future. Moreover, the communication may utilize any proper wireless communication technology, comprising but not limited to: Code Division Multiple Access (CDMA), Frequency Division Multiple Access (FDMA), Time Division Multiple Access (TDMA), Frequency Division Duplex (FDD), Time Division Duplex (TDD), Multiple-Input Multiple-Output (MIMO), Orthogonal Frequency Division Multiple (OFDM), Discrete Fourier Transform spread OFDM (DFT-s-OFDM) and / or any other technologies currently known or to be developed in the future.

[0060] Figure 2 illustrates an example of a control apparatus 200 for controlling a function of the access network (e.g., a 5G-RAN or the NG-RAN illustrated in Figures 1A and 1B) illustrated on Figures 1A and 1B. The control apparatus 200 may comprise at least one random access memory (RAM) 211 a, at least on read only memory (ROM) 211b, at least one processor 212, 213 and a network interface 214. The at least one processor 212, 213 may be coupled to the RAM 211a and the ROM 211b. The at least one processor 212, 213 may be configured to execute an appropriate software code 215. Execution of the software code 215 may for example may cause the apparatus to perform operations for controlling a function of the access network. The software code 215 may be stored in the ROM 211b. The control apparatus 200 may be interconnected with another control apparatus 200 for controlling another function of the 5G-RAN or the NG-RAN. In some embodiments, each function of the 5G-RAN or the NG-RAN is deployed or hosted on a control apparatus 200. In alternative embodiments, two or more functions of the 5G-RAN or the NG-RAN may share a control apparatus.

[0061] Figure 3 illustrates an example of a communication device 300, such as the UE illustrated on Figures 1A and 1B. The communication device 300 may be provided by any device capable of sending and receiving radio signals. Non-limiting examples of a communication device 300 comprise a user equipment, 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), a personal data assistant (PDA) or a tablet provided with wireless communication capabilities, a machine-type communications (MTC) device, an Internet of things (loT) type communication device or any combinations of these or the like. The communication device 300 may comprise a transceiver for transmitting and / or receiving, for example, wireless signals carrying communications, for example radio signals. The communications may be one or more of voice, electronic mail (email), text messages, multimedia data, machine data and so on.

[0062] The communication device 300 may receive wireless signals (e.g., radio signals) over an air or radio interface 307 via appropriate apparatus for receiving and may transmit wireless signals via appropriate apparatus for transmitting radio signals. In Figure 3 transceiver is designated schematically by block 306. The transceiver 306 may comprise, for example, a radio part and associated antenna arrangement. The antenna arrangement may be arranged internally or externally to the mobile device and may comprise one or more antenna elements. The antenna arrangement may be a multi-input multi-output (MIMO) antenna.

[0063] The communication device 300 may be provided with at least one processor 301, at least one memory ROM 302a, at least one RAM 302b 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 networks (e.g., the 5G-RAN or NG-RAN illustrated in Figures 1A and 1B) and other communication devices. The at least one processor 301 is coupled to the RAM 302b and the ROM 302a. The at least one processor 301 may be configured to execute an appropriate software code 308. The software code 308 may for example allow to perform one or more operations of the communication device. The software code 308 may be stored in the ROM 302a.

[0064] The processor, the ROM, and the RAM, the transceiver and other circuitry of the communication device (e.g., a modem) can be provided on a circuit board, in chipsets, or in a system on chip. The circuit board, chipsets or system on chip is denoted by reference 304. The communication device 300 may optionally have a user interface such as keypad 305, touch sensitive screen or pad, combinations thereof or the like. Optionally one or more of a display, a speaker and a microphone may be provided depending on the type of communication device.

[0065] Communication systems, such as described above in connection with Figures 1A to 3, are becoming increasingly data driven. In particular, they are increasing being configured to use cloud and distributed computing services to solve large-scale and complex problems for network configurations, network maintenance, and network optimization. Many use cases involve the collection of an extensive volume of data by data producers, and disseminating it to multiple data consumers, either periodically or reactively.

[0066] Different data consumers often have different, respective requirements for receiving and / or transmitting data (e.g., transmitting and / or receiving data at varying levels of granularity and within different time constraints). For example, the same Radio Access Network (RAN) data may need to be distributed via a Data Collection and Analysis Platform (DCAP) to various network automation applications, which may also be located at different physical servers. This may be the case, for example, for self-organizing networks (SON) solutions such as Mobility Robustness Optimization (MRO) and Mobility Load Balancing (MLB), or for synchronizing network digital twins.

[0067] In order to address limitations in communications available for transmission of such data, and limitations in storage and / or processing resources at the data producer and data consumer sides, representation models (RM) and generative models (GM) have been introduced.

[0068] Representation models offer a method for enabling data synthesist and compression. The aim of a representation model is to characterise a raw data set so that the characteristics of that raw data set can be determined by a data consumer.

[0069] In more detail, a data producer trains a representation model on data at the data producer's end and then transfers the representation model to the data consumer. The data producer trains the representation model (e.g., an autoencoder comprising an encoder and a decoder) to extract representations (e.g., characteristics) of raw data. When the data producer subsequently wants to transmit raw data to a data consumer, the data producer inputs the raw data to the encoder, which will result in an output of the representations of the raw data (with reduced dimension).

[0070] The data consumer can use the received representations in the following two alternative ways. First, the data consumer can use the representations as the input into other models or algorithms. For example, for a classification model to classify the network states, the representations can be used instead of the raw data. Similarly, for other prediction or optimization tasks, the representations can be used as the inputs instead of raw data. Second, when the data consumer decides to download the decoder of the representation model, then the data consumer can use the received representations as the input, and feed them into the decoder, such that the decoder can output the reconstructed data.

[0071] This is illustrated with reference to the example of Figure 4A to 4F.

[0072] Figure 4A illustrates an example in which the data producer 401A is comprised at a radio access network node. In the example of Figure 4A, the data producer inputs raw data into a representation model encoder, which outputs an encoded representation of the raw data. The encoded representation of the raw data is provided to a DCAP 402A (which is acting as a data consumer in this example) over a communication channel. The DCAP 402A inputs the received encoded representation into a representation model decoder, that outputs decoded data.

[0073] Generative models also offer a machine learning-based method for enabling data synthesis and compression. The aim of these models is to provide a method by which a raw data set can be reconstructed by a data consumer without the raw data set being transmitted by the data producer.

[0074] In more detail, generative models can be trained on data at the data producer's end and then be transferred to the data consumer. When a data producer subsequently wants to transmit raw data to a data consumer, the data producer determines what input to the generative model will result in the output of the raw data. The data consumer may subsequently receive that input for the generative model from the data producer, and cause the generative model to be used by inputting that input. The output of the generative model reconstructs the raw data, and is known as “synthetic data”).

[0075] A generative model may be configured to learn the data distribution and can generate random instances of observation X. Alternatively, the generative model may be a conditional generative model that learns the conditional probability P(X|Y) based on a given target or label Y. The recipient may use the received generative model to generate data that follows the same distribution as the original data.

[0076] Instead of signalling the generative model from the data producer to the data consumer, the data consumer may instead be provided with the output of the generative mode (e.g., with synthetic data). This latter case may be useful when an intermediary is used to transfer the requested data to the data consumer. For example, an intermediary may receive a generative model from a data producer, produce the synthetic data using the generative model, and provide the synthetic data to the data consumer. The synthetic data may be accompanied by information indicating (or otherwise identifying) the generative model used to create the synthetic data.

[0077] Stated differently, synthetic data generated by the generative model, along with optional information about the generative model, allows the data consumer to reproduce the generative model as it was used by the data producer. The data consumer may further combine the received synthetic data with its own local data in order to train a customized model located at the data consumer. The synthetic data may comprise information comparable to the original data, but may be smaller in size (e.g., compressed and / or with less noise and reduced bias) relative to the raw data.

[0078] Generative models and synthetic data transfers are illustrated with respect to figures 4B to 4C.

[0079] An example of transferring generative model between a DCAP and a data consumer is depicted in Figure 4B.

[0080] Figure 4B illustrates an example in which the data producer 401B is comprised at a DCAP. In the example of Figure 4B, the data producer inputs raw data into a learning agent, which outputs a generative model. The generative model is provided to a data consumer 402B over a communication channel. The data consumer 402B uses the received generative model to obtain the synthetic data.

[0081] Figure 4C illustrates an example in which synthetic data is transferred between a DCAP and a data consumer.

[0082] Figure 4C illustrates an example in which the data producer 401C is comprised at a DCAP. In the example of Figure 4C, the data producer inputs raw data into a learning agent, which outputs a generative model. The output of the generative model (e.g., the synthetic data) is subsequently provided to a data consumer 402C over a communication channel, as is information indicating the generative model used for creating the synthetic data. The data consumer 402C subsequently uses the synthetic data as the requested data.

[0083] As an aside, it is noted that, in the present disclosure, the term “data consumer” may simply refer to an entity (e.g., an apparatus) that requests (and subsequently receives) data from another entity (e.g., the “data producer”). The data producer may not be a source that originates the data. Stated differently, the data producer may be an entity between a data consumer and a data source. Further, the data consumer may not be the ultimate end point of the requested data. Stated differently, the data consumer may act as a data producer to another entity. Some examples of different types of data consumers and data producers are illustrated with respect to Figures 4D to4F.

[0084] Figure 4D illustrates signalling that may be performed between a user equipment 401D, a cloud server 402D, and a network access node 403D. It is understood that each of these entities may comprise a constraint aware delivery service (CADS) function that causes the signalling of Figure 4D to be performed.

[0085] During 4001D, the user equipment 401D signals the cloud server 402D. The signalling of 4001D may comprise a request for data (“the requested data”).

[0086] During 4002D, the cloud server 402D signals the network access node 403D. This signalling of 4002D may comprise a request for data to be received from the network access node for fulfilling the request of 4001D. It is understood that, although not shown, the cloud server 402D may signal other entities during 4002D for obtaining data that will be used in fulfilling the data request of 4001D.

[0087] During 4003D, the network access node 403D signals the cloud server 402D. This signalling of 4003D may comprise the data requested in the data request of 4002D.

[0088] During 4004D, the cloud server 402D signals the user equipment 401D. This signalling of 4004D may comprise the data requested in the data request of 4001D.

[0089] Depending on the signalling used and data transferred, different entities in Figure 4D may be considered as being data consumers and data producers. In this example, the user equipment 401D may be considered a data consumer of the cloud server 402D, and / or a data consumer of the network access node 403D. Further, the cloud data consumer 402D may be considered to be a data consumer of the network access node 403D, and / or a data producer of the user equipment 401D. Further, the network access node 403D may be considered to be a data producer of the cloud server 402D, and / or a data producer of the user equipment 401D.

[0090] Figure 4E illustrates an example of signalling between a network access node 401E and a user equipment 402E. In the example of Figure 4E, the network access node 401E is considered to act as a data consumer to the user equipment 402E, and the user equipment 402E is configured to act as a data producer to the network access node 401E.

[0091] During 4001E, the network access node 401E signals the UE 402E. This signalling of 4001E comprises a request for data to be received from the UE.

[0092] During 4002E, the UE 402E signals the network access node 401E. This signalling of 4002E may comprise the data requested during 4001E.

[0093] Figure 4F illustrates an example of signalling between a user equipment 401F and a network access node 402F. In the example of Figure 4F, the user equipment 401F is considered to act as a data consumer to the network access node 402F, and the network access node 402F is configured to act as a data producer to the user equipment 401F.

[0094] During 4001F, the user equipment 401F signals the network access node 402F. This signalling of 4001F comprises a request for data to be received from the UE.

[0095] During 4002F, the network access node 402F signals the user equipment 401F. This signalling of 4002F may comprise the data requested during 4001F.

[0096] By using representation models and / or generative models, the volume of data that needs to be transmitted for providing a data consumer with a raw data set can be significantly reduced, thus conserving transmission resources. Furthermore, this approach can alleviate the storage burden on both data producers and data consumers as generative models can generate high-fidelity approximations of the raw data set, which obviates the need to store and transmit large raw datasets.

[0097] Therefore, there are at least four different formats in which data is provided from a data producer to a data consumer. These are: • Raw data: data corresponding to data requirements requested by a data consumer. • Representation: the representation may be provided alone or with the representation model. • Generative model: • Synthetic data: the synthetic data may optionally be accompanied by model information that identifies the generative model used for creating the synthetic data.

[0098] The use of such generative models in communication systems are also gaining attention in standardization efforts. For example, during the 3GPP SA WG2 Meeting #152E, a working item focusing on “using generative models for data storage and transfer” was introduced. This item allows network functions (such as, for example, a Network Data Analytics Function (NWDAF) and / or an Analytics Data Repository Function (ADRF)) to use data synthesis and compression techniques to generate input data for inference or training, which reduces the volume of data that needs to be transferred and stored.

[0099] However, due to the dynamic nature of network data, the accuracy of generative models may fluctuate over time. Moreover, communicational constraints between data producers and data consumers pose additional challenges. In more detail, data consumers may face difficulties in receiving or processing data within the required data transfer delay limits, leading to the age of information expiring and the data becoming outdated and less useful. Additionally, computational limits on hard disk storage, random access memory (RAM), and processing capability on the data consumer side further complicate the transfer and employment of generative models.

[0100] The following describes a method for addressing at least one of the above-mentioned issues.

[0101] In more detail, the following proposes at least one method in which a functional entity (labelled herein as a Constraints-Aware Data Delivery Service (CADS)) of a data producer and / or a data consumer recommends and / or decides a most appropriate data format of information delivery among the raw data, representations, generative model, and synthetic data.

[0102] The most appropriate data format may be a data format that is selected based on at least one constraint for performing the selection (e.g., at least one policy, criteria, threshold, etc.).

[0103] For example, the most appropriate data format may be selected using at least one parameter relating to one or more of the following considerations, which are described further above: « Data transfer requirements • Communication capacity • Computational requirements • Computational capacity • Model quality

[0104] Moreover, the selection of the data format does not need to be static. Instead, the CADS may cause the dynamic switching between different information formats over time. The switching may be performed based on changes in the at least one parameter. Whether a data format is to be switched from one form to another form may be evaluated periodically, and / or aperiodically (e.g., in response to a threshold associated with a parameter being breeched, which in turn triggers a switch and / or an evaluation for whether to perform a switch being run).

[0105] By dynamically selecting the information formats, a constraints-aware data delivery service may enable energy and resource efficient data sharing, optimizing both communication and computational resource utilization while meeting the needs of diverse data consumers on both data fidelity and time sensibility.

[0106] These principles will be further illustrated with respect to Figures 5 to 9. In more detail, Figure 5 provides an illustration of an example system in which the presently described techniques may be implemented (in addition to the above examples of Figures 4D to 4F), Figures 6 to 7 illustrate example signalling between apparatus of Figure 5, and Figures 8 and 9 illustrate example methods that may be performed by data consumers and data producers of the provided examples. There are also provided more specific examples that are not illustrated by Figures.

[0107] An overview of a system in which the presently described techniques may be illustrated is shown in Figure 5.

[0108] Figure 5 illustrates a plurality of data consumers 501A to 501D, each of which are configured to request data from a data producer 502 (e.g., a DCAP) using respective requests. The request for data may be unique per data consumer. Stated differently, the request of data for a data consumer may be provided independently of any other request for data transmitted by another data consumer. The data producer 502 collects data from a plurality of different sources 503A, 503B. The data producer evaluates, for each data request, which data format should be used for responding to the data request, and selects a data format for that data request based on that evaluation. This evaluation may be performed using the at least one parameter mentioned above. Based on the evaluation, the data producer provides the requested data in the form of one of, raw data, representation, a generative model, and synthetic data (as described above).

[0109] At least one of the parameters known by the data consumer may be provided to the data producer from the data consumer. This may be performed using the same signalling as the data request, and / or may be performed using alternative signal as the data request.

[0110] For example, each data request may identify parameters (e.g., at least one of the data requirements and / or constraints) known by the data consumer. For example, a data request may identify at least one of, for example, data requirements such as a data type, data characteristics, an estimated amount of data, a data transfer delay time, as well as information on data consumer’s computation capabilities and constraints. Alternatively, the CADS data consumer may configure the CADS data producer according to data requirements and constraints.

[0111] The data producer may further be configured to provide the data consumer an indication of a list of the different types of data formats available to the data consumer, along with an indication of an expected performance of each type of data format. The data consumer may use this received list and indicated expected performance to select a data format, and indicate the selected data format to the data producer. The data producer may use the selected data format received from the data consumer as the data format for provided the requested data to the data consumer. In a variation to this, the data consumer may, in response to receiving the list of the different types of data formats available to the data consumer, along with an indication of an expected performance of each type of data format, associate a respective priority for each type of data format, and provide the data producer with the respective priorities. In this case, the data producer may use the respective priorities for selecting the data format as the data format for provided the requested data to the data consumer. The provision of respective priorities may be useful for dynamic reselection of the data format.

[0112] The above-mentioned concepts are illustrated with respect to the following examples of Figures 6 to 9. In more detail, Figures 6 to 7 illustrate example signalling that may be performed between apparatus in order to illustrate various concepts, while Figures 8 and 9 illustrate functionality of the examples of Figures 6 to 7. It is therefore understood that, where the terminology used between these Figures are the same, the discussion in respect of Figures 6 to 7 (and in the rest of the present application) may be used to better understand functionality described in Figures 8 to 9.

[0113] The examples of Figures 6 and 7 relate to examples in which a data consumer requests data to be provided from a data producer.

[0114] Figure 6 illustrates a first example in which a data consumer 601 communicates with a data producer 602.

[0115] During 6001, the data consumer 601 signals the data producer 601. This signalling may comprise a request for data to be received from the data producer 602.

[0116] The signalling of 6001 may comprise one or more data requirements, wherein the data requirements specify at least one quality and / or characteristic to be fulfilled by the requested data. These requirements may comprise, for example, details such as the data type, time granularity, time duration, and other characteristics such as throughput and delay requirements for the data transfer. It is also possible to include more specific requirements, such as requesting a specific format for being provided with the requested data (e.g., raw data only).

[0117] During 6002, the data producer 602 collects the requested raw data and stores the raw data (e.g., by writing into (temporary) reports or logs). If the data producer is able to, the data producer may also use the collected raw data to train a representation model and / or a generative model.

[0118] During 6003, the data consumer 601 and data producer 602 exchange signalling for determining a current (and / or estimating a future) condition of the communication channel between the data consumer 601 and the data producer 602 over which the requested data is to be provided.

[0119] Stated differently, during 6003, the data producer 602 evaluates a communication channel for the requested data transfer between the data source and recipient, e.g., estimating the achievable throughput and delay on the data transfer channel. This step may comprise both the data consumer and the data producer (e.g., because pilot or feedback signals may be involved to measure the throughput and delay in the downlink direction). However, it is understood that this the channel condition evaluation of 6003 is not limited to being performed by exchanging communications between the data producer 602 and the data consumer 601 (e.g., the data producer 602 may determine the communication channel condition(s) autonomously).

[0120] During 6004, the data producer 602 estimates computational requirements for each of the different data formats. For example, during 6004, the data producer 602 may estimate computing requirements to store the raw data, to store the representations, to store the synthetic data, to execute the representation model, and / or to execute the generative mode. This estimation may consider, for example, processing and memory resources used for this purpose by, for example, at least one of a CPU, a GPU, hard disk storage, or temporary memory.

[0121] During 6005, the data producer 602 estimates a respective model quality associated with each of the representation model and the generative model. Stated differently, the data producer estimates whether the output of the different models is as expected (e.g., that the output is accurate within a preconfigured degree of tolerance).

[0122] For example, during 6005, the apparatus may determine the accuracy of RM and GM by detecting any data drift from each of the models to detect whether the RMs and GMs meet the requirements of the accuracy and fidelity.

[0123] When it is determined that a particular model is “inaccurate” (e.g., because the data drift is more than a preconfigured threshold), that model may be finetuned or retrained until it becomes “accurate” again. Stated differently, when the measured accuracy of a model is less than a first preconfigured threshold (e.g., less than 90% accurate), then that model may be retrained. It is understood that the level of the first preconfigured threshold may be set by the data requirements of the data consumer (e.g., as communicated during 6001).

[0124] It is also understood that when the accuracy of a model is very high, and the requested data does not require that high a level of accuracy, then some of the data that is well represented by the model (e.g., such as older data) can be removed to release space in the hard disk drive. Stated differently, when the measured accuracy of a model exceeds a second preconfigured threshold (e.g., 99% or more), the data producer causes at least some data that is well represented in that model to be deleted. It is understood that the level of the second preconfigured threshold may be set by the data requirements of the data consumer (e.g., as communicated during 6001).

[0125] During 6006, the data producer 602 creates an indication that summarises the different available data formats for providing the requested data of 6001 along with at least one evaluation score or metric for each of the available data formats that indicates a relative quality of that data format.

[0126] This is illustrated with respect to Table 1, which shows an example of the summarized information of different possible formats and their corresponding 5 requirements and evaluations. It is understood that this is merely used as an example, and that different types and formats of the requirements and quality measures can be included in the summary instead of the information in Table 1. For example, the quality does not need to be a score between 0 and 100. Moreover, the provided information can include detail metrics to evaluate the quality of the synthetic data or models. 10 Alternatively an even coarser mapping may be used that it linked to limited performance classes (e.g., the three classes good, satisfying, poor). information format Network requiremen ts to transfer data Computational requirements to store / execute the data / representations / model Quality of Data / Representatio ns / Model CPU GPU Hard disk storag e RA M Raw data Throughput 100 MB / s, Delay 1 second for 100 MB report per second N / A N / A 8.64 TB / da y N / A 100 Representatio ns Throughput 5 MB / s, Delay 1 second for 5 MB report per second N / A N / A 432 GB / da y N / A 80 Representatio ns with decoder Throughput 5 MB / s, Delay 1 second for 5 MB report per second 64 bit process or Above GeForc e GT 730, AMD Radeo n R7 240 432 GB / da y + 50 MB 2 GB 80 GM N / A 64 bit process or Above GeForc e GT 730, AMD Radeo n R7 240 100 MB 4 GB 95 Synthetic data Throughput 10 MB / s, Delay 1 second for 10 MB report per second N / A N / A 864 GB / da y N / A 95 Table 1 Example of summarised description of different formats

[0127] During 6007, the data producer 602 signals the data consumer 601. The signalling of 6007 may cause the data consumer to be provided with the indication that summarises the different available data formats for providing the requested data of 6006. For example, during 6007, the data producer may send the data consumer summarised information (such as the information in Table 1) about the requirements for using the different formats and respective quality of the different data formats. It is understood that the data producer may send information about all of the different data formats, or may send information about only a subset (e.g., less than all) of the different data formats. In this latter case, the subset may be selected based on at least one requirement and / or quality criteria to be met by the requested data. The at least one requirement and / or quality criteria may be comprised in the signalling of 6001, and / or be preconfigured at the data producer during some previous configuration operation.

[0128] During 6008, the data consumer 601 assesses a computational capacity of the data consumer. The assessed computational capacity may be for determining which (if any) of the indicated data formats may be able to be processed by the data consumer over an estimated time scale for processing that data format. For example, the data consumer 601 may assess computational capacity such as, for example, CPU resources, GPU resources, hard disk storage resources, and RAM resources for processing the different data formats. The computational capacity may be computational capacity that is local to the data consumer 601.

[0129] During 6009, the data consumer selects at least one of the data formats. The selection may be performed based on the results of 6008 and / or any requirements for the requested data. Stated differently, the data consumer may select an “optimal” data format for the information to be delivered to the data consumer based on one or more of: the assessed computational capacity, and the received summary indicating data requirements, computational requirements, or the evaluated model and data quality.

[0130] The determination of which data format is an “optimal” data format may consider one of more of a plurality of different factors and / or use any of a plurality of different methods.

[0131] For example, one possible formulation of selecting an optimal data forma may be to minimize the data transfer among data sources (e.g., RANs), DCAPs, and data consumers, subjected to a set of the constraints listed in below: o Network capacities for data transfer (including achievable bandwidth and delay) that are to be satisfied the data transfer requirements (including data throughput and delay constraints) for transferring the requested data in that data format from the data producer to the data consumer. o Hardware computing capacity of the data consumer, including the CPU, GPU, hard disk storage, and temporary memory, and whether the computing capacity can satisfy the hosting and / or storage of the transferred data / representations / model. o The quality and fidelity of representations, RM, GM, and / or synthetic data relative to the requirements of the data request.

[0132] The selection of 6009 may select a single one of the data formats. When the selection of 6010 selects a plurality of the data formats, the data consumer may associate each of the plurality of the data formats with a respective priority that indicates a preference for using that data format. Stated differently, where a plurality of data formats is selected by the data consumer, the data consumer may determine a ranking that ranks each of the data formats within the plurality of data formats for selection.

[0133] During 6010, the data consumer 601 signals the data producer 602. The signalling of 6010 may comprise an indication of the at least one selected data format of 6009. The signalling of 6010 may comprise an indication of a single data format. The signalling of 6010 may comprise an indication of a plurality of data formats. As mentioned above, the indication of the plurality of data formats may be accompanies by a ranking that indicates a priority (and / or preference) for selecting each of the data formats relative to the other data formats in the plurality of data formats.

[0134] During 6011, the data producer 602 uses the indication of 6010 to select a final data format for providing the requested data of 6001 to the data consumer from the at least one selected data format received by the data producer during 602. Where only one data format is indicated during 6010, the selection of 6011 selects that one data format as the final data format.

[0135] During 6012, the data producer 602 signals the data consumer 601. This signalling may comprise the requested data in the final data format of 6011.

[0136] Figure 7 illustrates another example method in which the presently described principles may be illustrated. The example of Figure 7 differs from the example of Figure 6 in that the data producer of Figure 7 uses the above-mentioned computational capacity of the data consumer instead of the data consumer of Figure 6.

[0137] Figure 7 illustrates a first example in which a data consumer 701 communicates with a data producer 702.

[0138] During 7001, the data consumer 701 signals the data producer 701. This signalling may comprise a request for data to be received from the data producer 702.

[0139] The signalling of 7001 may comprise one or more data requirements, wherein the data requirements specify at least one quality and / or characteristic to be fulfilled by the requested data. These requirements may comprise, for example, details such as the data type, time granularity, time duration, and other characteristics such as throughput and delay requirements for the data transfer. It is also possible to include more specific requirements, such as requesting a specific format for being provided with the requested data (e.g., raw data only). 7001 may be as described above in relation to 6001.

[0140] During 7002, the data producer 702 collects the requested raw data and stores the raw data (e.g., by writing into (temporary) reports or logs). If the data producer is able to, the data producer may also use the collected raw data to train a representation model and / or a generative model. 7002 may be as described above in relation to 6002.

[0141] During 7003, the data consumer 701 and data producer 702 exchange signalling for determining a current (and / or estimating a future) condition of the communication channel between the data consumer 701 and the data producer 702 over which the requested data is to be provided.

[0142] Stated differently, during 7003, the data producer 702 evaluates a communication channel for the requested data transfer between the data source and recipient, e.g., estimating the achievable throughput and delay on the data transfer channel. This step may comprise both the data consumer and the data producer (e.g., because pilot or feedback signals may be involved to measure the throughput and delay in the downlink direction). However, it is understood that this the channel condition evaluation of 7003 is not limited to being performed by exchanging communications between the data producer 702 and the data consumer 701 (e.g., the data producer 702 may determine the communication channel condition(s) autonomously).

[0143] 7003 may be as described above in relation to 6003.

[0144] During 7004, the data producer 702 estimates computational requirements for each of the different data formats. For example, during 7004, the data producer 702 may estimate computing requirements to store the raw data, to store the representations, to store the synthetic data, to execute the representation model, and / or to execute the generative mode. This estimation may consider, for example, processing and memory resources used for this purpose by, for example, at least one of a CPU, a GPU, hard disk storage, or temporary memory. 7004 may be as described above in relation to 6004.

[0145] During 7005, the data producer 702 estimates a respective model quality associated with each of the representation model and the generative model. Stated differently, the data producer estimates whether the output of the different models is as expected (e.g., that the output is accurate within a preconfigured degree of tolerance).

[0146] 7005 may be as described above in relation to 6005.

[0147] During 7006, the data producer 702 creates an indication that summarises the different available data formats for providing the requested data of 7001 along with at least one evaluation score or metric for each of the available data formats that indicates a relative quality of that data format.

[0148] 7003 may be as described above in relation to 6006.

[0149] During 7007, the data producer 702 signals the data consumer 701. The signalling of 7007 may be performed when at least one of the representation model and / or the generative model is determined to produce an output that has an accuracy and / or fidelity that fulfils at least one criteria to be used by the data consumer 701.

[0150] During 7008, the data consumer 701 assesses a computational capacity of the data consumer. For example, the data consumer 701 may assess computational capacity such as, for example, CPU resources, GPU resources, hard disk storage resources, and RAM resources for processing the different data formats. The computational capacity may be computational capacity that is local to the data consumer 701.

[0151] During 7009, the data consumer 701 signals the data producer 702. The signalling of 7009 may provide the data producer with an indication of the computational capacity of the data consumer 701 from 7008.

[0152] During 7010, the data producer selects an “optimal” data format for the information to be delivered to the data consumer based on one or more of: the received computational capacity, and a similar summary indicating data requirements, computational requirements, or the evaluated model and data quality as described above in relation to 7006 and 6006.

[0153] The determination of which data format is an “optimal” data format may as described above in relation to 6009 (except that in 7010, this is performed by the data producer 702 instead of the data consumer 602).

[0154] During 7011, the data producer 702 signals the data consumer 701. This signalling may comprise the requested data in the final data format of 7011.

[0155] The example of Figure 7 may be advantageous over the example of Figure 6 when a data consumer later requests the summarisation information described above, in that it allows a data consumer to expose its computational constraints to the data producer, such that the data producer can screen the candidate information formats and provides a better, cleaner summary of suitable data formats. This may also reduce the size of any the summary information to be sent to data recipient in the event the data consumer later requests the summary information.

[0156] Although it is understood that the presently described methods may be implemented in a range of different types of communication networks, the following presents an example of some of the considerations of implementing the presently described methods in a 3GPP 5G (and / or beyond, such as 6G) network communication environment. It is understood that the following is not limiting, but may be used to provide greater understanding of the above-described principles and methods. Subsequent to this, Figures 8 to 9 will highlight a plurality of features and functionalities of the present disclosure.

[0157] As discussed above, 3GPP describes scenarios in which a large amount of data needs to be collected and shared with multiple data consumers. This may be performed a plurality of times (e.g., periodically), with different granularity and possibly with different time constraints for the different data consumers and / or at different times. For example, the same RAN data (say mobile trace data on Massive MIMO and grid of beams) may need to be shared with different network automation applications, with network digital twins or with service management functions. In mobility trace data example, the geolocation and mobility-related data can be logged with various time granularities, ranging from seconds to minutes. Assuming the basic mobility trace data record per user per second is around 100 bytes, the data collected per user per day amounts to 8.64 MB. For a network with 1 million users, this results in 8.64 TB of data per day to be transmitted and stored. Transmitting and storing such raw data is resource and energy consuming. However, there are limited communication resources to send the data to the respective data consumers. Further, there are different computational constraints or the memory spaces at either the data consumer or data producer to aggregate the data.

[0158] When the bandwidth and memory constraints do not allow the delivery of raw unprocessed data, generative machine learning models can be used for data synthesis and compression, where the generative models are trained in a data producer and transferred to a data consumer to reduce bandwidth and memory costs required to transfer data. However, the data can only be regenerated at the data consumer using a generative model if sufficient computational capabilities are available at the data consumer for running that model. Finally, the quality of generated data may need to fulfil preconfigured data accuracy and fidelity requirements of the data consumer. If this is not the case, low-dimensional representations of the data may be more suitable.

[0159] However, it may not be desired to exchange representations or GMs, e.g., due to data privacy concerns (in the case of representations) or proprietary intellectual property concerns (in the case of GMs), Instead, synthetic data derived for the GMs may be delivered.

[0160] When and which data format is the most suitable to be transmitted to the data consumer depends on various factors and needs to be carefully estimated. A constraints-aware data delivery service (e.g., a functionality comprised in apparatus described herein) can evaluate the required data and applicable constraints and decide the appropriate data format for delivering the data (e.g., either the raw data, representations, a generative model that can be used to synthesize the data, or synthetic derived from the GM).

[0161] To address at least one of these issues in 3GPP (which may also be issues in other networks), a 3GPP management system (MnS) may be configured according to the present disclosure to have a capability enabling an MnS data consumer (e.g., an entity that requests data from the MnS) to request for data with an indication of a format in which the data may be delivered, either as raw data, as representations, as a network-data generative model or synthetic data derived using the network-data generative model.

[0162] Further, the MnS may be configured to have a capability to deliver requested data to an MnS data consumer either as raw data, as representations, as a networkdata generative model or synthetic data derived using the network-data generative model.

[0163] The 3GPP management system may further be configured to have a capability for enabling the MnS data consumer to configure a data producer with the MnS data consumer’s data-related requirements and constraints (e.g., via a policy) including communication and computation constraints, as well as network-data generative model performance requirements.

[0164] The 3GPP management system may further be configured to have a caoabiltiy for enabling the MnS data consumer to request and receive information on the performance of a generative network-data model that can be provided by the MnS data producer in place of raw data produced by the Mns data producer.

[0165] To help effect this, the data producer Information Object Class (IOC) described in current 3GPP specifications (e.g., in 3GPP TS 28.622) may be modified to additionally comprise an attribute for indicating the supported data formats. The data producer IOC class may represent the function that generates network-related data. This data producer IOC may be considered as an enumeration with the values as: {RAW_DATA, REPRESENTATION, GENERATIVE_MODEL, SYNTHETIC_DATA}

[0166] The data producer IOC may be provided to (and read by) the data MnS data consumer to understand the alternative data formats that may be provided by the MnS data producer. The data producer may be requested by the MnS data consumer provide data either as raw data, representations, generative model that can be used to synthesize the desired data or as synthetic data derived using the network-data generative model.

[0167] For example, the data producer IOC may be as illustrated below in Table 2. Attribute name Support Qualifie r isReada ble isWrita ble islnvaria nt isNotifya ble rawData Mandate ry True False False True representationM odel Mandato ry True False False True generativeModel Optional True False False True syntheticData Optional True False False True Table 2 New Data Producer IOC attribute

[0168] In this example of Table 2, there is provided: * an attribute for raw data: • an attribute for a representation that contains the knowledge in the data but in reduced dimension, e.g., using an autoencoder; * an attribute for generative model for data synthesis, which would carry a network-data generative model trained to contain the knowledge contained in a large data set. This attribute may indicate to the data consumer that a generative model can be provided by the MnS data producer instead of the MnS data producer raw data. As mentioned above, the data content may also contain the attribute for the performance of network; ® an attribute for synthetic data that contains the synthetic data that has been generated by a generative ML model.

[0169] The data producer IOC may further comprise an attribute indicating the performance of network-data generative model. This can be read by the data MnS data consumer to understand the expected performance of the generative model. A potential form for this is indicated below in Table 3. Attribute name isReadabl e isWritabl e islnvaria nt isNotifyabl e supportedDataFormats True False True True dataGenModelPerforman ce True False True True Table 3 New Data Producer IOC indicating generative model performance

[0170] As discussed above, Figures 8 to 9 illustrate methods that may be performed by apparatus discussed herein. It is therefore understood that description provided above may be used to better understand the terms and functionality mentioned below.

[0171] Figure 8 illustrates a method that may be implemented by an apparatus of a data producer. It is understood in the examples of Figures 8 and 9 that the terms “providing” and “transmitting” are used synonymously and interchangeably, and that the terms “obtaining” and “receiving” are used synonymously and interchangeably.

[0172] During 8001, the data producer provides, to a data consumer, an attribute indicating, for each of a plurality of data formats, whether the apparatus is able to support providing data in said data format, wherein the plurality of different data formats comprises: raw data, a representation of the raw data, a generative model for regenerating the raw data as synthetic data, and the synthetic data.

[0173] Stated differently, during 8001, the apparatus of Figure 8 provides at least one indication to the data consumer that indicates, for each of four different data formats ( raw data, a representation of the raw data, a generative model for regenerating the raw data as synthetic data, and the synthetic data), whether (or not) the apparatus is configured to be able to provide data to the data consumer in that data format. For example, the apparatus may be configured to be able to provide data to the data consumer in all of the four formats. As another example, the apparatus may be configured to be able to provide data to the data consumer in a subset (e.g., less than all) of the four data formats.

[0174] The attribute (e.g., indication) may be explicit in the signalling of 8001. Stated differently, for each of the four data formats, information may be comprised in the attribute by comprising a respective value in one or more fields of the signalling of 8001 that denotes the availability of that specific data format for providing information between the data consumer and the data producer.

[0175] The raw data may be the data requested in the request for data. The raw data may comprise data collected by the data producer for responding to the request of 8001.

[0176] The representation of the raw data may comprise an encoded form of the raw data that, when decoded, indicates characteristics of the raw data and / or a reconstructed form of the raw data. The representation may be formed using a representation model, such as described above.

[0177] The generative model may be a model that, when executed, outputs the synthetic data, wherein the synthetic data is as described below.

[0178] The synthetic data may approximate the raw data within a predetermined accuracy threshold. Stated differently, the synthetic data may be said to reproduce the raw data (within a predetermined accuracy). Stated differently, the synthetic data may be said to replicate the raw data (within a predetermined accuracy). Stated differently, the synthetic data may be said to duplicate the raw data (within a predetermined accuracy). Stated differently, the synthetic data may be a recreated form of the raw data (within a predetermined accuracy).

[0179] During 802, the apparatus obtains, from the data consumer, a request for data to be provided to the data consumer using at least one of said plurality of data formats, the at least one of said plurality of data formats comprising a first data format.

[0180] The request may be provided in any of a plurality of different forms.

[0181] For example, the request may be provided in the form of conditions and / or constraints that indicates which of the different data formats is preferred in different contexts (e.g., with different computing and / or storage resources available at the different entities, with different channel conditions between the data consumer and the data producer, and / or with different traffic requirements for the requested data). The different contexts and / or constraints may be comprised in a policy.

[0182] The request may indicate a single data format (e.g., the first data format). The request may indicate all of the plurality of data formats. The request may indicate less than all of the plurality of data formats (e.g., 1,2 or 3 data formats).

[0183] During 803, the apparatus provides the requested data to the data consumer using the first data format.

[0184] The apparatus of Figure 8 may select the first data format from the at least one of said data formats.

[0185] It is understood that although the following refers to the selection of the first data format in a singular step, that the data producer / apparatus of Figure 8 may repeatedly perform a selection of a data format from the plurality of data formats for providing requested data to a data consumer. This is especially useful when the data request is provided in the form of a subscription request that requests the delivery of data on a plurality of different transmission occasions. In such a case, the reselection may be performed in the same manner as the initial selection, although it is understood that changes in at least one of the factors mentioned below may lead to a different data format being selected than the first data format.

[0186] The selection of the first data format may be performed by considering at least one of a plurality of different factors.

[0187] For example, one of the factors may comprise a state (e.g., a quality) of a communication channel over which the requested data is to be provided. In such a case, the apparatus of Figure 8 may evaluate a current and / or estimated property of a communication channel over which the requested data is to be provided, wherein the selecting the first data format is based on the current and / or estimated property and / or wherein the attribute is configured based on the current and / or estimated property. The property may comprise at least one of a signal quality (e.g., a reference signal received power, a signal-to-interference-and-noise ratio, etc.), a channel capacity, a channel throughput, or an available bandwidth. The apparatus may receive an indication of the current and / or estimated property from the data consumer. The received indication of the signal property may have previously been requested from the data consumer.

[0188] As another factor for the selecting, the apparatus of Figure 8 may determine, for each of the four data formats, what computing resources (e.g., processing resources and / or storage resources) are used when that data format is used for providing data to the data consumer. For example, the apparatus of Figure 8 may estimate computing requirements for performing two or more of the following at the data consumer and / or at the data producer: storing the raw data, storing the representation of raw data, generating the representation of raw data using an encoder, generating the synthetic data, storing a decoder of the representation, and / or storing the generative model. As mentioned above, the selecting the first data format is based on the estimated computing requirements and / or wherein the attribute is configured based on the estimated computing requirements.

[0189] Another factor in the selecting may include a preference for at least one different data format that is determined by the data consumer. For example, the apparatus of Figure 8 may provide, to the data consumer, an indication of the estimated computing requirements for two or more data formats of said plurality of data formats, obtain (from the data consumer) an indication of a preferred data format for said providing the requested data, and select the first data format based on the indicated preferred data format. As mentioned above, the preference may be provided in the form of a policy that comprises a plurality of constraints and / or conditions, and / or the preference may simply comprise a ranking that ranks one of the data formats over the other data formats.

[0190] The apparatus of Figure 8 may further obtain, from the data consumer, an indication of a computing capacity of the data consumer, and the selecting may use the indicated computing capacity and the estimated computing requirements to select first data format, and / or to provide, to the data consumer, an indication of the estimated computing requirements for two or more of said plurality of data formats.

[0191] Another factor in the selecting may comprise a consideration of an accuracy and / or fidelity of the representation model and / or generative model.

[0192] To assist in this, the apparatus of Figure 8 may track an accuracy and / or fidelity of at least one of the generative model or a representation model that forms the representation of raw data over time (e.g., by comparing the respective outputs of the models to their intended output). Stated differently, the apparatus of Figure 8 may track respective data drifts corresponding to each model, and determine whether the data drift exceeds a predetermined threshold. As mentioned above, the selecting the first data format may be based on the tracked accuracy and / or fidelity, and / or the attribute may be configured based on the tracked accuracy and / or fidelity.

[0193] When the model is not accurate enough to be used as a data format for providing the data consumer with requested data, the model may be modified until it is accurate enough. For example, the tracking the accuracy and / or fidelity may comprise detecting whether the generative model and / or the representation model produces an output that fulfils a first predetermined criteria of accuracy and / or a first predetermined criteria of fidelity. When it is detected that the generative model and / or the representation model does not fulfil the first predetermined criteria of accuracy and / or the first predetermined criteria of fidelity, the apparatus may modify the generative model and / or the representation model until the first predetermined criteria of accuracy and / or the first predetermined criteria of fidelity is fulfilled.

[0194] Conversely, when the model is very accurate (e.g., more than a threshold amount than needed by associated data requirements of the requested data), the apparatus may save some storage resources by removing older and / or well represented data from the model. Stated differently, when it is determined that the generative model and / or the representation model fulfils a second predetermined criteria of accuracy and / or a second predetermined criteria of fidelity, the apparatus may remove older (e.g., well represented in the model) data from the generative model and / or the encoder (e.g., the representation model).

[0195] The apparatus of Figure 8 may further collect the raw data, and use the collected raw data to train at least one of the generative model and / or a representation model that outputs the representation data.

[0196] Optionally, when the first format is the representation of raw data, the apparatus of Figure 8 may determine whether to provide a decoder of the raw data to the data consumer, and provide the decoder to the data consumer when it is determined to provide the decoder. The decoder may decode an input representation to output the characteristics of the raw data or the reconstructed data itself, as discussed herein.

[0197] Figure 9 illustrates a method that may be performed by an apparatus for a data consumer. The data consumer of Figure 9 may be the data consumer of Figure 8. The data consumer of Figure 9 may be considered an interacting apparatus to the data producer of Figure 8. it is understood in the following that terms and features mentioned below that are mentioned in relation to claim 8 may comprise the same functionality and meaning as described in connection with claim 8.

[0198] During 901, the apparatus obtains, from a data producer, an attribute indicating, for each of a plurality of data formats, whether the data producer is able to support providing data in said data format, wherein the plurality of different data formats comprises: raw data, a representation of the raw data, a generative model for regenerating the raw data as synthetic data, and the synthetic data. This may be as described above in relation to 801.

[0199] During 902, the apparatus provides, to the data producer, a request for data to be provided to the apparatus using at least one of said plurality of data formats, the at least one of said plurality of data formats comprising a first data format. This may be as described in connection with 802.

[0200] During 903, the apparatus obtains the requested data using the first data format. This may be as described above in connection with 803.

[0201] The apparatus of Figure 9 may, before said obtaining the data during 903, obtain, from the data producer, an indication of estimated computing requirements for two or more of said forms for obtaining said data, select a preferred form for obtaining said data based on the estimated computing requirements and a current and / or estimated future computing capacity of the apparatus, and provide an indication of the preferred form to the data producer. The selection of the preferred form may be as described above in connection with Figure 8.

[0202] The apparatus of Figure 9 may provide, to the data producer, an indication of a computing capacity of the apparatus. This may be as described above in connection with Figure 8.

[0203] The apparatus of Figure 9 may, when the first format is the representation of raw data, obtain a decoder for decoding the representation data from the data producer.

[0204] The following may apply in respect of the apparatus of Figures 8 and 9.

[0205] The requested data may be accompanied by an indication that identifies the first data format. For example, the requested data of 803 and / or 804 may be comprised in signalling that further comprises a field that enumerates what the first data format is (e.g., raw data, representation data, synthetic data, or a generative model).

[0206] The attribute may be provided with an indication that indicates an expected performance of the generative model. The expected performance may relate to an accuracy and / or fidelity of the generative model. This may be as described herein.

[0207] The request for data may comprise at least one policy that defines at least one constraint and / or condition for selecting the first data format from the plurality of data formats.

[0208] For example, the request may comprise at least one constraint for selecting the first data format from the plurality of data formats, the at least one constraint comprising at least one of: a data type, time granularity, time duration, throughput requirements for data transfer, delay requirements for data transfer, or a request for a specific form for receiving the data.

[0209] It is understood that references in the above to various network functions (e.g., to an AMF, an SMF, TNF etc.) may comprise apparatus that perform at least some of the functionality associated with those network functions. Further, an apparatus comprising a network function may comprise a virtual network function instance of that network function.

[0210] It should be understood that the apparatuses may comprise or be coupled to other units or modules etc., such as radio parts or radio heads, used in or for transmission and / or reception. Although the apparatuses have been described as one entity, different modules and memory may be implemented in one or more physical or logical entities.

[0211] It is noted that whilst some embodiments have been described in relation to 5G networks, similar principles can be applied in relation to other networks and communication systems. Therefore, although certain embodiments were described above by way of example with reference to certain example architectures for wireless networks, technologies and standards, embodiments may be applied to any other suitable forms of communication systems than those illustrated and described herein.

[0212] It is also noted herein that while the above describes example embodiments, there are several variations and modifications which may be made to the disclosed solution without departing from the scope of the present invention.

[0213] As used herein, “at least one of the following: ” and “at least one of ” and similar wording, where the list of two or more elements are joined by “and” or “or”, mean at least any one of the elements, or at least any two or more of the elements, or at least all the elements.

[0214] In general, the various embodiments may be implemented in hardware or special purpose circuitry, software, logic or any combination thereof. Some aspects of the disclosure 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 the disclosure is not limited thereto. While various aspects of the disclosure 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.

[0215] As used herein, the term “circuitry” may refer to one or more or all of the following: (a) hardware-only circuit implementations (such as implementations in only analog and / or digital circuitry) and (b) combinations of hardware circuits and software, such as (as applicable): (i) a combination of analog 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 a mobile phone or server, to perform various functions) 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.”

[0216] This definition of circuitry applies to all uses of this term herein, including in any claims. As a further example, as used herein, 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 and if applicable to the particular claim element, a baseband integrated circuit or processor integrated circuit for a mobile device or a similar integrated circuit in server, a cellular network device, or other computing or network device.

[0217] The embodiments of this disclosure may be implemented by computer software executable by a data processor of the mobile device, such as in the processor entity, or by hardware, or by a combination of software and hardware. Computer software or program, also called program product, including software routines, applets and / or macros, may be stored in any apparatus-readable data storage medium and they comprise program instructions to perform particular tasks. A computer program product may comprise one or more computer-executable components which, when the program is run, are configured to carry out embodiments. The one or more computer-executable components may be at least one software code or portions of it.

[0218] Further in this regard it should be noted that any blocks of the logic flow as in the Figures 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. The physical media is a non-transitory media.

[0219] The term “non-transitory,” as used herein, is a limitation of the medium itself (i e., tangible, not a signal) as opposed to a limitation on data storage persistency (e.g , RAM vs. ROM).

[0220] 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 comprise one or more of general purpose computers, special purpose computers, microprocessors, digital signal processors (DSPs), application specific integrated circuits (ASIC), FPGA, gate level circuits and processors based on multi core processor architecture, as non-limiting examples.

[0221] Various example embodiments of the disclosure may be practiced in various components such as integrated circuit modules. The design of integrated circuits is by and large a highly automated process. Complex and powerful software tools are available for converting a logic level design into a semiconductor circuit design ready to be etched and formed on a semiconductor substrate.

[0222] The scope of protection sought for various example embodiments of the disclosure is set out by the independent claims. The example embodiments and features thereof, if any, described in this disclosure that do not fall under the scope of the independent claims are to be interpreted as examples useful for understanding various example embodiments of the disclosure.

[0223] The foregoing description has provided, by way of non-limiting and illustrative examples, a full and informative description of the various example embodiments of this disclosure. 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 claims. However, all such and similar modifications of the teachings will still fall within the various example embodiments of the disclosure as set forth in the claims. By way of non-limiting and illustrative example, there is a further example embodiment comprising a combination of one or more example embodiments with any of the other example embodiments previously discussed.

Claims

1) An apparatus comprising means for:providing, to a data consumer, an attribute indicating, for each of a plurality of data formats, whether the apparatus is able to support providing data in said data format, wherein the plurality of different data formats comprises: raw data, a representation of the raw data, a generative model for regenerating the raw data as synthetic data, and the synthetic data;obtaining, from the data consumer, a request for data to be provided to the data consumer using at least one of said plurality of data formats, the at least one of said plurality of data formats comprising a first data format; andproviding the requested data to the data consumer using the first data format.2) An apparatus as claimed in claim 1, wherein the apparatus further comprises means for selecting the first data format from the at least one of said data formats.3) An apparatus as claimed in claim 2, the apparatus further comprising means for: evaluating a current and / or estimated property of a communication channel over which the requested data is to be provided, wherein the selecting the first data format is based on the current and / or estimated property and / or wherein the attribute is configured based on the current and / or estimated property, wherein the property comprises at least one of a signal quality, a channel capacity, a channel throughput, or an available bandwidth.4) An apparatus as claimed in claim 3, further comprising means for receiving an indication of the current and / or estimated property from the data consumer.5) An apparatus as claimed in any of claims 2 to 4, further comprising means for estimating computing requirements for performing two or more of: storing the raw data, storing the representation of raw data, generating the representation of raw data using an encoder, generating the synthetic data, storing a decoder of the representation, and / or storing the generative model, wherein theselecting the first data format is based on the estimated computing requirements and / or wherein the attribute is configured based on the estimated computing requirements.6) An apparatus as claimed in claim 5, wherein the selecting means further comprises means for:providing, to the data consumer, an indication of the estimated computing requirements for two or more data formats of said plurality of data formats;obtaining, from the data consumer, an indication of a preferred data format for said providing the requested data; andselecting the first data format based on the indicated preferred data format.7) An apparatus as claimed in any of claims 5 to 6, wherein the selecting means further comprises means for: obtaining, from the data consumer, an indication of a computing capacity of the data consumer, and wherein the means for selecting uses the indicated computing capacity and the estimated computing requirements to select first data format, and / or to provide, to the data consumer, an indication of the estimated computing requirements for two or more of said plurality of data formats.8) An apparatus as claimed in any of claims 2 to 7, further comprising means for: tracking an accuracy and / or fidelity of at least one of the generative model or a representation model that forms the representation of raw data over time, wherein the means for selecting comprises means for selecting the first data format based on the tracked accuracy and / or wherein the attribute is configured based on the tracked accuracy.9) An apparatus as claimed in claim 8, wherein the means for tracking the accuracy and / or fidelity comprises means for detecting whether the generative model and / or the representation model produces an output that fulfils a first predetermined criteria of accuracy and / or a first predetermined criteria offidelity, and the apparatus further comprising means for, when it is detected that the generative model and / or the representation model does not fulfil the first predetermined criteria of accuracy and / or the first predetermined criteria of fidelity, modifying the generative model and / or the representation model until the first predetermined criteria of accuracy and / or the first predetermined criteria of fidelity is fulfilled.10)An apparatus as claimed in any of claims 8 to 9, further comprising means for, when it is determined that the generative model and / or the representation model fulfils a second predetermined criteria of accuracy and / or a second predetermined criteria of fidelity, removing older data from the generative model and / or the representation model.11 )An apparatus as claimed in any preceding claim, further comprising means for: collecting the raw data; andtraining at least one of the generative model and / or a representation model that outputs the representation data using the collected raw data.12)An apparatus as claimed in any preceding claim, further comprising means for, when the first format is the representation of raw data: determining whether to provide a decoder of the raw data to the data consumer, and providing the decoder to the data consumer when it is determined to provide the decoder.13)An apparatus comprising means for:obtaining, from a data producer, an attribute indicating, for each of a plurality of data formats, whether the data producer is able to support providing data in said data format, wherein the plurality of different data formats comprises: raw data, a representation of the raw data, a generative model for regenerating the raw data as synthetic data, and the synthetic data;providing, to the data producer, a request for data to be provided to the apparatus using at least one of said plurality of data formats, the at least one of said plurality of data formats comprising a first data format; andobtaining the requested data using the first data format.14)An apparatus as claimed in claim 13, further comprising means for, before said obtaining the data:obtaining, from the data producer, an indication of estimated computing requirements for two or more of said forms for obtaining said data;selecting a preferred form for obtaining said data based on the estimated computing requirements and a current and / or estimated future computing capacity of the apparatus; andproviding an indication of the preferred form to the data producer.15)An apparatus as claimed in any of claims 13 to 14, further comprising means for providing, to the data producer, an indication of a computing capacity of the apparatus.16)An apparatus as claimed in any of claims 13 to 14, further comprising means for, when the first format is the representation of raw data: obtaining a decoder for decoding the representation data from the data producer.17)An apparatus as claimed in any preceding claim, wherein the requested data is accompanied by an indication that identifies the first data format.18)An apparatus as claimed in any preceding claim, wherein the attribute is provided with an indication that indicates an expected performance of the generative model.19)An apparatus as claimed in any preceding claim, wherein the request for data comprises at least one policy that defines at least one constraint and / or condition for selecting the first data format from the plurality of data formats.20)An apparatus as claimed in claim 19, wherein the request comprises at least one constraint for selecting the first data format from the plurality of data formats, the at least one constraint comprising at least one of: a data type, time granularity, time duration, throughput requirements for data transfer, delayrequirements for data transfer, or a request for a specific form for receiving the data.21)An apparatus as claimed in any preceding claim, wherein:the raw data is the data requested in the request for data;the representation of the raw data comprises an encoded form of the raw data that, when decoded, indicates characteristics of the raw data and / or a reconstructed form of the raw data;the generative model is a model that, when executed, outputs the synthetic data, wherein the synthetic data approximates the raw data within a predetermined accuracy threshold; and / orthe synthetic data approximates the raw data within a predetermined accuracy threshold.22)A method for an apparatus, the method:providing, to a data consumer, an attribute indicating, for each of a plurality of data formats, whether the apparatus is able to support providing data in said data format, wherein the plurality of different data formats comprises: raw data, a representation of the raw data, a generative model for regenerating the raw data as synthetic data, and the synthetic data;obtaining, from the data consumer, a request for data to be provided to the data consumer using at least one of said plurality of data formats, the at least one of said plurality of data formats comprising a first data format; andproviding the requested data to the data consumer using the first data format.23)A computer program that, when executed by a computer of an apparatus, causes the apparatus to perform:providing, to a data consumer, an attribute indicating, for each of a plurality of data formats, whether the apparatus is able to support providing data in said data format, wherein the plurality of different data formats comprises: raw data, a representation of the raw data, a generative model for regenerating the raw data as synthetic data, and the synthetic data;obtaining, from the data consumer, a request for data to be provided to the data consumer using at least one of said plurality of data formats, the at least one of said plurality of data formats comprising a first data format; andproviding the requested data to the data consumer using the first data format.24)A method for an apparatus, the method comprising:obtaining, from a data producer, an attribute indicating, for each of a plurality of data formats, whether the data producer is able to support providing data in said data format, wherein the plurality of different data formats comprises: raw data, a representation of the raw data, a generative model for regenerating the raw data as synthetic data, and the synthetic data;providing, to the data producer, a request for data to be provided to the apparatus using at least one of said plurality of data formats, the at least one of said plurality of data formats comprising a first data format; andobtaining the requested data using the first data format.25)A computer program that, when executed by a computer of an apparatus, causes the apparatus to perform:obtaining, from a data producer, an attribute indicating, for each of a plurality of data formats, whether the data producer is able to support providing data in said data format, wherein the plurality of different data formats comprises: raw data, a representation of the raw data, a generative model for regenerating the raw data as synthetic data, and the synthetic data;providing, to the data producer, a request for data to be provided to the apparatus using at least one of said plurality of data formats, the at least one of said plurality of data formats comprising a first data format; andobtaining the requested data using the first data format.Application No: GB2411118.9Examiner:Adam TuckerClaims searched: 1-25Date of search: 10 December 2024Patents Act 1977: Search Report under Section 17Documents considered to be relevant:Category Relevant to claims Identity of document and passage or figure of particular relevance X A,E A A A A 1-25 WO 2021 / 018370 Al (Ericsson) See the whole document and in particular Figures 1, 9, paragraphs 26-36 WO 2024 / 211535 Al (Convida Wireless, LLC) See the whole document and in particular paragraphs 48, 63-65, 82, 101, 113 (Table 4), 134, 135, 195-196 US 2024 / 0056837 Al (Nokia Solutions and Networks Oy) See the whole document and in particular paragraphs 13, 50, 71, 75-78, 91, 92, 160-161 WO 2024 / 027916 Al (Nokia Technologies Oy) See the whole document and in particular paragraphs 69, 76-79 WO 2023 / 087247 Al (Huawei Technologies Co. Ltd.) See the whole document WO 2023 / 220948 Al (Huawei Technologies Co. Ltd.) See the whole documentCategories: X Document indicating lack of novelty or inventive step A Document indicating technological background and / or state of the art. Y Document indicating lack of inventive step if combined with one or more other documents of same category. P Document published on or after the declared priority date but before the filing date of this invention. & Member of the same patent family E Patent document published on or after, but with priority date earlier than, the filing date of this application.Field of Search:Search of GB, EP, WO &US patent documents classified in the following areas of the UKCX :Worldwide search of patent documents classified in the following areas of the IPC____________G06F; G06N; H03M; H04L; H04W________________________________The following online and other databases have been used in the preparation of this search report SEARCH-PATENT, SEARCH-NPLwww.gov.uk / ipoInternational Classification:Subclass Subgroup Valid From H04L 0069 / 04 01 / 01 / 2022 G06N 0003 / 0455 01 / 01 / 2023 G06N 0003 / 0475 01 / 01 / 2023 H03M 0007 / 30 01 / 01 / 2006 H04L 0041 / 16 01 / 01 / 2022 H04L 0069 / 08 01 / 01 / 2022 H04L 0069 / 24 01 / 01 / 2022 H04W 0024 / 02 01 / 01 / 2009 H04L 0067 / 04 01 / 01 / 2022 H04L 0067 / 5651 01 / 01 / 2022 H04L 0067 / 60 01 / 01 / 2022 H04W 0008 / 22 01 / 01 / 2009www.gov.uk / ipo

Images