Training data collection

The system transforms vendor-specific training data into a unified format for ML models, addressing data imbalance and enhancing network performance by diversifying and augmenting training datasets across different vendors.

JP7954285B2Active Publication Date: 2026-11-19NOKIA TECHNOLOGIES OY
View PDF 4 Cites 0 Cited by

Patent Information

Application Number
JP2025518579
Authority / Receiving Office
JP · JP
Patent Type
Patents
Current Assignee / Owner
Priority Date
2022-09-28
Filing Date
2023-08-16
Publication Date
2026-11-19
Estimated Expiration
2043-08-16

AI Technical Summary

Technical Problem

Existing techniques for collecting and processing training data in wireless telecommunication networks often result in unexpected outcomes due to the use of vendor-specific data, leading to unbalanced and suboptimal training datasets for Machine Learning models.

Method used

A system comprising a coordinator function that configures collector functions to collect and transform training data into a vendor-independent format, combining it with data from different vendors to create a unified training dataset for ML models, using operations like superposition, averaging, and filtering.

Benefits of technology

This approach ensures robust and vendor-independent ML model training by diversifying and augmenting training data, resulting in improved network performance and reduced reliance on proprietary data sharing.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 0007954285000001
    Figure 0007954285000001
  • Figure 0007954285000002
    Figure 0007954285000002
  • Figure 0007954285000003
    Figure 0007954285000003
Patent Text Reader

Abstract

1. An apparatus comprising: a coordinator function configured to send a first configuration message to a first collector function, the first configuration message including information for configuring the first collector function to collect first training data, transform the first training data to generate transformed first training data, and report the transformed first training data.
Need to check novelty before this filing date? Find Prior Art

Description

TECHNICAL FIELD

[0001] Various example embodiments relate to collecting training data. BACKGROUND ART

[0002] In wireless telecommunication networks, network nodes perform collection and processing of training data to train Machine Learning (ML) models to improve network operation. While existing techniques for collecting and processing training data exist, unexpected results may occur. Accordingly, it is desirable to provide improved techniques for collecting and processing training data. SUMMARY OF THE INVENTION

[0003] The scope of protection sought for various example embodiments of the present invention is set out by the independent claims. Any example embodiments and features described in this specification that do not fall within the scope of the independent claims, if present, should be construed as useful examples for understanding various embodiments of the present invention.

[0004] According to various example embodiments of the present invention, not necessarily all of which, there is provided an apparatus comprising a coordinator function configured to send a first configuration message to a first collector function, wherein the first configuration message comprises information for configuring the first collector function to collect first training data, transform the first training data to generate transformed first training data, and report the transformed first training data.

[0005] The first training data may comprise an N-dimensional matrix of values collected by the first collector function.

[0006] The value may include channel information. This channel information may relate to the wireless link between the entity hosting the collector function and the transmitter. The channel information may include beamforming and / or channel values.

[0007] The first configuration message may include information for configuring the first collector function to transform the first training data to produce the transformed first training data by removing specified vendor-specific data and / or artifacts. The specified vendor-specific data and / or artifacts may include radio frequency receiver delays, several radio frequency chains, and similar.

[0008] The first configuration message may include information for configuring the first collector function to transform the first training data to generate transformed first training data with target reconstruction.

[0009] The first configuration message may include information to configure the first collector function to report the converted first training data in the converted training data format and / or at a specified reporting interval.

[0010] The transformed first training data can include an N-dimensional matrix of values ​​collected by the first collector function.

[0011] The first configuration message may include information to configure the first collector function to report the converted first training data to the coordinator function and / or training function.

[0012] The first configuration message may include information to configure the first collector function to report the transformed first training data to both the first and second training functions.

[0013] The coordinator function may be configured to send a second configuration message to a second collector function, the second configuration message containing information for configuring the second collector function to collect second training data, reconstruct the second training data to produce transformed second training data, and report the transformed second training data.

[0014] The coordinator function may be configured to combine the received transformed training data to form combined transformed training data. The received transformed training data may come from multiple collector functions and / or from the same collector function at different times.

[0015] The coordinator function may be configured to combine the received transformed training data by performing superposition, averaging, filtering, pruning, puncturing, scaling, and / or normalization.

[0016] The coordinator function may be configured to send the combined and transformed training data to the training function.

[0017] The coordinator function may be configured to send configuration messages to a collector function provided by a vendor common to both the coordinator and collector functions, the configuration messages including instructions to configure the collector function to collect training data and report the training data.

[0018] The coordinator function may be configured to send a conversion configuration message to the training function, which includes information for configuring the training function to convert the received converted training data and / or received combined converted training data into converted training data.

[0019] The conversion configuration message may include information for configuring the training function to convert the received transformed training data and / or the received combined transformed training data into converted training data by performing expansion, overlay, averaging, filtering, pruning, puncturing, normalization, scaling, and / or translation. The conversion may also be performed by appropriate filtering or projection operations.

[0020] The coordinator function may be configured to send a first conversion configuration message to a first training function, the first conversion configuration message including information for configuring the first training function to convert the received converted training data and / or received combined converted training data into first converted training data, and to send a second conversion configuration message to a second training function, the second conversion configuration message including information for configuring the second training function to convert the received converted training data and / or received combined converted training data into second converted training data.

[0021] Messages and / or data may be transmitted over physical sidelink shared channels, physical downlink shared channels, physical uplink shared channels, and / or any wireless medium channels. Messages and / or data may be transmitted over physical sidelink control channels, physical downlink common control channels, and / or physical uplink common channels. Messages and / or data may be transmitted over non-air interfaces, such as F1, Xn, and / or NG interfaces. Messages and / or data may be transmitted over O-RAN A1, E1, E2, and F1 interfaces. It will be understood that these are applicable to 4G, 5G, and 6G systems.

[0022] The training data, transformed training data, and / or transformed training data may include a data format specified by a combination of parameters associated with the training data, such as sampling resolution, array shape and / or length, such as scalar, vector, or matrix, and similar.

[0023] According to various exemplary embodiments, not necessarily all, of the present invention, a method is provided which includes sending a first configuration message to a first collector function, the first configuration message containing information for configuring the first collector function to collect first training data, transform the first training data to generate transformed first training data, and report the transformed first training data.

[0024] The first training data can include an N-dimensional matrix of values ​​collected by the first collector function.

[0025] The value may include channel information. This channel information may relate to the wireless link between the entity hosting the collector function and the transmitter. The channel information may include beamforming and / or channel values.

[0026] The first configuration message may include information for configuring the first collector function to transform the first training data to produce the transformed first training data by removing specified vendor-specific data and / or artifacts. The specified vendor-specific data and / or artifacts may include radio frequency receiver delays, several radio frequency chains, and similar.

[0027] The first configuration message may include information for configuring the first collector function to transform the first training data to generate transformed first training data with target reconstruction.

[0028] The first configuration message may include information for configuring the first collector function to report the converted first training data in a converted training data format and / or at a specified reporting period.

[0029] The converted first training data may include an N-dimensional matrix of values collected by the first collector function.

[0030] The first configuration message may include information for configuring the first collector function to report the converted first training data to a coordinator function and / or a training function.

[0031] The first configuration message may include information for configuring the first collector function to report the converted first training data to both the first training function and the second training function.

[0032] The method may include sending a second configuration message to a second collector function, where the second configuration message includes information for configuring the second collector function to collect second training data, reconstruct the second training data to generate converted second training data, and report the converted second training data.

[0033] The method may include combining the received converted training data to form combined converted training data. The received converted training data may be from multiple collector functions, and / or from the same collector function at different times.

[0034] The method may include combining the received converted training data by performing superimposition, averaging, filtering, pruning, puncturing, scaling, and / or normalization.

[0035] The method could include sending the combined and transformed training data to the training function.

[0036] Sending may be performed by a coordinator function, and the method may include sending a configuration message to a collector function provided by a vendor common to the coordinator function and the collector function, wherein the configuration message includes instructions for configuring the collector function to collect training data and report the training data.

[0037] The method may include sending a conversion configuration message to the training function, the conversion configuration message containing information for configuring the training function to convert the received converted training data and / or received combined converted training data into converted training data.

[0038] The conversion configuration message may include information for configuring the training function to convert the received transformed training data and / or the received combined transformed training data into converted training data by performing expansion, overlay, averaging, filtering, pruning, puncturing, normalization, scaling, and / or translation. The conversion may also be performed by appropriate filtering or projection operations.

[0039] The method may include sending a first conversion configuration message to a first training function, the first conversion configuration message including information for configuring the first training function to convert the received converted training data and / or received combined converted training data into first converted training data; and sending a second conversion configuration message to a second training function, the second conversion configuration message including information for configuring the second training function to convert the received converted training data and / or received combined converted training data into second converted training data.

[0040] Messages and / or data may be transmitted over physical sidelink shared channels, physical downlink shared channels, physical uplink shared channels, and / or any wireless medium channels. Messages and / or data may be transmitted over physical sidelink control channels, physical downlink common control channels, and / or physical uplink common channels. Messages and / or data may be transmitted over non-air interfaces, such as F1, Xn, and / or NG interfaces. Messages and / or data may be transmitted over O-RAN A1, E1, E2, and F1 interfaces. It will be understood that these are applicable to 4G, 5G, and 6G systems.

[0041] The training data, transformed training data, and / or transformed training data may include a data format specified by a combination of parameters associated with the training data, such as sampling resolution, array shape and / or length, such as scalar, vector, or matrix, and similar.

[0042] According to various exemplary embodiments of the present invention, not necessarily all, an apparatus is provided comprising at least one processor and at least one memory storing instructions that, when executed by the at least one processor, cause the apparatus to carry out at least the methods and exemplary embodiments described above.

[0043] According to various exemplary embodiments of the present invention, not necessarily all, a non-temporary computer-readable medium is provided which includes stored program instructions for carrying out the methods and exemplary embodiments described above.

[0044] According to various embodiments of the present invention, not necessarily all, the present invention provides a collector function configured to receive configuration messages from a coordinator function, wherein the configuration messages include information for configuring the collector function to collect training data, transform the training data to generate transformed training data, and report the transformed training data.

[0045] The training data can include an N-dimensional matrix of values ​​collected by a collector function.

[0046] The value may include channel information. This channel information may relate to the wireless link between the entity hosting the collector function and the transmitter. The channel information may include beamforming and / or channel values.

[0047] The configuration message may include information to configure the collector function to transform the training data and generate transformed training data by removing specified vendor-specific data and / or artifacts. Specified vendor-specific data and / or artifacts may include radio frequency receiver delays, several radio frequency chains, and similar.

[0048] The configuration message may include information to configure the collector function to transform the training data and generate transformed training data with target reconstruction.

[0049] The configuration message may include information to configure the collector function to report the converted training data in the converted training data format and / or at a specified reporting interval.

[0050] The transformed training data can include an N-dimensional matrix of values ​​collected by the collector function.

[0051] The configuration message may include information to configure the collector function to report the converted training data to the coordinator function and / or the training function.

[0052] The configuration message may include information to configure the collector function to report the converted training data to both the first and second training functions.

[0053] The collector function may be configured to receive configuration messages from a coordinator function provided by a vendor common to the collector function, and the configuration messages include instructions for configuring the collector function to collect training data and report the training data.

[0054] Messages and / or data may be transmitted over physical sidelink shared channels, physical downlink shared channels, physical uplink shared channels, and / or any wireless medium channels. Messages and / or data may be transmitted over physical sidelink control channels, physical downlink common control channels, and / or physical uplink common channels. Messages and / or data may be transmitted over non-air interfaces, such as F1, Xn, and / or NG interfaces. Messages and / or data may be transmitted over O-RAN A1, E1, E2, and F1 interfaces. It will be understood that these are applicable to 4G, 5G, and 6G systems.

[0055] The training data, transformed training data, and / or transformed training data may include a data format specified by a combination of parameters associated with the training data, such as sampling resolution, array shape and / or length, such as scalar, vector, or matrix, and similar.

[0056] According to various embodiments of the present invention, not necessarily all, the present invention provides a method for receiving a configuration message from a coordinator function, the configuration message containing information for configuring a collector function to collect training data, transform the training data to generate transformed training data, and report the transformed training data.

[0057] The training data can include an N-dimensional matrix of values ​​collected by a collector function.

[0058] The value may include channel information. This channel information may relate to the wireless link between the entity hosting the collector function and the transmitter. The channel information may include beamforming and / or channel values.

[0059] The configuration message may include information to configure the collector function to transform the training data and generate transformed training data by removing specified vendor-specific data and / or artifacts. Specified vendor-specific data and / or artifacts may include radio frequency receiver delays, several radio frequency chains, and similar.

[0060] The configuration message may include information to configure the collector function to transform the training data and generate transformed training data with target reconstruction.

[0061] The configuration message may include information to configure the collector function to report the converted training data in the converted training data format and / or at a specified reporting interval.

[0062] The transformed training data can include an N-dimensional matrix of values ​​collected by the collector function.

[0063] The configuration message may include information to configure the collector function to report the converted training data to the coordinator function and / or the training function.

[0064] The configuration message may include information to configure the collector function to report the converted training data to both the first and second training functions.

[0065] The method may include receiving a configuration message from a coordinator function provided by a common vendor to the collector function, the configuration message containing instructions for configuring the collector function to collect training data and report the training data.

[0066] Messages and / or data may be transmitted over physical sidelink shared channels, physical downlink shared channels, physical uplink shared channels, and / or any wireless medium channels. Messages and / or data may be transmitted over physical sidelink control channels, physical downlink common control channels, and / or physical uplink common channels. Messages and / or data may be transmitted over non-air interfaces, such as F1, Xn, and / or NG interfaces. Messages and / or data may be transmitted over O-RAN A1, E1, E2, and F1 interfaces. It will be understood that these are applicable to 4G, 5G, and 6G systems.

[0067] The training data, transformed training data, and / or transformed training data may include a data format specified by a combination of parameters associated with the training data, such as sampling resolution, array shape and / or length, such as scalar, vector, or matrix, and similar.

[0068] According to various exemplary embodiments of the present invention, not necessarily all, an apparatus is provided comprising at least one processor and at least one memory storing instructions that, when executed by the at least one processor, cause the apparatus to carry out at least the methods and exemplary embodiments described above.

[0069] According to various exemplary embodiments of the present invention, not necessarily all, a non-temporary computer-readable medium is provided which includes stored program instructions for carrying out the methods and exemplary embodiments described above.

[0070] According to various embodiments of the present invention, not necessarily all, the present invention provides a training function configured to receive a conversion configuration message from a coordinator function, wherein the conversion configuration message includes information for configuring the training function to convert the received converted training data and / or received combined converted training data into converted training data.

[0071] The conversion configuration message may include information for configuring the training function to convert the received transformed training data and / or the received combined transformed training data into converted training data by performing expansion, overlay, averaging, filtering, pruning, puncturing, normalization, scaling, and / or translation. The conversion may also be performed by appropriate filtering or projection operations.

[0072] The training data can include an N-dimensional matrix of values ​​collected by a collector function.

[0073] The value may include channel information. This channel information may relate to the wireless link between the entity hosting the collector function and the transmitter. The channel information may include beamforming and / or channel values.

[0074] The transformed first training data can include an N-dimensional matrix of values ​​collected by the collector function.

[0075] The training function may be configured to combine the received transformed training data to form combined transformed training data.

[0076] The training function may be configured to combine the received transformed training data by performing superposition, averaging, filtering, pruning, puncturing, scaling, and / or normalization.

[0077] Messages and / or data may be transmitted over physical sidelink shared channels, physical downlink shared channels, physical uplink shared channels, and / or any wireless medium channels. Messages and / or data may be transmitted over physical sidelink control channels, physical downlink common control channels, and / or physical uplink common channels. Messages and / or data may be transmitted over non-air interfaces, such as F1, Xn, and / or NG interfaces. Messages and / or data may be transmitted over O-RAN A1, E1, E2, and F1 interfaces. It will be understood that these are applicable to 4G, 5G, and 6G systems.

[0078] The training data, transformed training data, and / or transformed training data may include a data format specified by a combination of parameters associated with the training data, such as sampling resolution, array shape and / or length, such as scalar, vector, or matrix, and similar.

[0079] According to various embodiments of the present invention, not necessarily all, the present invention provides a method for receiving a conversion configuration message from a coordinator function, wherein the conversion configuration message includes information for configuring a training function to convert the received converted training data and / or received combined converted training data into converted training data.

[0080] The conversion configuration message may include information for configuring the training function to convert the received transformed training data and / or the received combined transformed training data into converted training data by performing expansion, overlay, averaging, filtering, pruning, puncturing, normalization, scaling, and / or translation. The conversion may also be performed by appropriate filtering or projection operations.

[0081] The training data can include an N-dimensional matrix of values ​​collected by a collector function.

[0082] The value may include channel information. This channel information may relate to the wireless link between the entity hosting the collector function and the transmitter. The channel information may include beamforming and / or channel values.

[0083] The transformed first training data can include an N-dimensional matrix of values ​​collected by the collector function.

[0084] The method may include combining the received transformed training data to form combined transformed training data.

[0085] The method may include combining the received transformed training data by performing superposition, averaging, filtering, pruning, puncturing, scaling, and / or normalization.

[0086] Messages and / or data may be transmitted over physical sidelink shared channels, physical downlink shared channels, physical uplink shared channels, and / or any wireless medium channels. Messages and / or data may be transmitted over physical sidelink control channels, physical downlink common control channels, and / or physical uplink common channels. Messages and / or data may be transmitted over non-air interfaces, such as F1, Xn, and / or NG interfaces. Messages and / or data may be transmitted over O-RAN A1, E1, E2, and F1 interfaces. It will be understood that these are applicable to 4G, 5G, and 6G systems.

[0087] The training data, transformed training data, and / or transformed training data may include a data format specified by a combination of parameters associated with the training data, such as sampling resolution, array shape and / or length, such as scalar, vector, or matrix, and similar.

[0088] According to various exemplary embodiments of the present invention, not necessarily all, an apparatus is provided comprising at least one processor and at least one memory storing instructions that, when executed by the at least one processor, cause the apparatus to carry out at least the methods and exemplary embodiments described above.

[0089] According to various exemplary embodiments of the present invention, not necessarily all, a non-temporary computer-readable medium is provided which includes stored program instructions for carrying out the methods and exemplary embodiments described above.

[0090] Further specific and preferred embodiments are described in the attached independent and dependent claims. Features of the dependent claims may be combined with features of the independent claims as necessary, and may be in combinations other than those expressly described in the claims.

[0091] When a feature of a device is described as being operable to perform a function, it will be understood that the feature of the device includes the features of the device that provide this function, or the features of the device that are adapted or configured to perform this function.

[0092] Some examples of embodiments will be described below with reference to the attached drawings. [Brief explanation of the drawing]

[0093] [Figure 1] This is a diagram showing insufficient training data. [Figure 2] This is a signal chart where data from Vendor 2 is used to enhance data from Vendor 1. [Figure 3] This is a signal chart in which data from Vendor 2 is used to enhance data from Vendor 1, and the controller function (NR-C) performs the integration of vendor-independent (agnostic) / domain-invariant data (DID1 and DID2). [Figure 4] This is a signal chart when multiple NR_Es report DID to multiple NR_Ts—this approach can be combined with the approach in Figure 2 and / or Figure 3. [Figure 5] This is a signal chart when multiple NR_Es report DID to multiple NR_Ts—this approach can be combined with the approaches in Figures 2, 3, and / or 4. [Figure 6] This is a signal chart for an example of SL positioning. [Figure 7] This is a diagram of an enhanced positioning use case. The entries in the DVD (or DID) matrix, DVD(a,b), represent UE-specific positioning measurements acquired in frequency resource a*Fs and time resource b*Ts, where Fs and Ts are the UE-specific sampling frequency and sampling time, respectively. [Modes for carrying out the invention]

[0094] Before discussing the embodiments of the examples in more detail, an overview is first provided. Some embodiments of the examples provide a technique in which functions for coordinating, collecting, and using training data to train ML models to perform various network and / or device-specific tasks are provided to network nodes in a wireless telecommunications network, commonly referred to as radio resource management (RRM). Typically, a collection function within the network node, where the training function that uses the training data is provided by the same vendor as the network node, can receive the training data along with its values ​​and in a format known to this training function. In some embodiments, even the collection function within the network node, where the training function that uses the training data is provided by the same vendor as the network node, provides the training data in an independent or immutable form. A collector function within the network node, where the training function that uses the training data is provided by a different vendor than the network node, provides the training data in an independent or immutable form that does not disclose vendor-specific information about the capabilities of the entity that collected the data. This training data can be provided to a coordinator function that combines the received data, or to a training function for combining the received data. The training function typically provides information such as transformation details, allowing the combined data to be subsequently transformed or converted into a format that can then be used by the training function to train its (vendor-specific) ML model. This approach helps collect a diverse range of training data in a consistent manner from network nodes provided by other vendors.

[0095] Some of the examples relate to the Study Item (SI) [3GPP RP-213599] for Artificial Intelligence (AI) / Machine Learning (ML) for the Rel-18 New Radio (NR) Air Interface. The SI aims to explore the benefits of extending the air interface with features that enable support for AI / ML-based algorithms for enhanced performance and / or reduced complexity / overhead. The goal of this SI is to lay the foundation for future air interface use cases to leverage AI / ML techniques. An initial set of use cases to be covered includes channel state information (CSI) feedback enhancement (e.g., overhead reduction, improved accuracy, prediction, etc.), beam management (e.g., beam prediction in the time and / or spatial domains for overhead and latency reduction, improved beam selection accuracy, etc.), positioning accuracy enhancement, and similar. For these use cases, the benefits must be evaluated (using developed methods and defined key performance indicators (KPIs)), and the potential impact on the specifications, including physical (PHY) layer aspects, protocol aspects, etc., must be assessed. One of the main expected consequences of the SI is that "AI / ML approaches for selected sub-use cases must be diverse enough to support a variety of requirements at the gNode B user equipment (gNB-UE) collaboration level." It should be noted that other use cases may also be addressed as an addition during the Work Item (WI) phase of "AI / ML for Air Interfaces." Starting with Release 18, it is very likely that a wide variety of use cases and applications for ML in gNBs and UEs will be proposed. Rel.17 Positioning Reference Unit (PRU) - A PRU is a 5G network entity that can be designated by a 5G NR network to support one or more positioning sessions.A PRU is a known-location device or network node (e.g., a roadside unit, another UE, etc.) that can be activated on demand by a location management function (LMF) to perform specific positioning operations, such as measuring and / or transmitting a unique positioning signal. RAN1 Liaison Document (LS) [R2-2106920] on PRUs - RAN1 evaluates the use of known-location positioning reference units (PRUs) for positioning and observes improvements in the use of PRUs to enhance positioning performance. Note: The term “positioning reference unit (PRU)” is used solely as technical terminology in this discussion. A PRU does not necessarily imply the introduction of a new network node. A PRU can, if agreed, support at least some of the UE’s Rel-16 positioning capabilities, which is at the discretion of RAN2. Positioning functions may include, but are not limited to, providing positioning measurements (e.g., reference signal time difference (RSTD), reference signal received power (RSRP), receive-transmit (Rx-Tx) time difference), and transmitting uplink (UL) sounding reference signals (SRS) for positioning. A PRU may be required by the LMF to provide the LMF with its own known location coordinate information. If the PRU's antenna orientation information is known, this information may also be required by the LMF. Rel. 18 It is expected that ML models for Radio Resource Management (RRM) will be vendor-specific and therefore trained with vendor-specific data. A foreseeable consequence of RAN is that companies agree that vendor-specific ML models will be trained for the same RRM functions using only vendor-specific training data, so that training data is not exchanged between vendors. This is one of several reasons why vendors (UEs and / or gNBs) may not want to share that data. In other words, data is unique to the user entity (UE), often confidential, and gives vendors a competitive advantage by enabling them to generate and deploy ML-based solutions that outperform competitors' solutions.

[0096] As illustrated in Figure 1, data collection for training vendor-specific ML models is therefore expected to be a lengthy process, most likely resulting in an unbalanced, suboptimal training dataset that remains large in imbalance between minority and majority labels, and is sparse, meaning the collected data does not adequately characterize all scenarios of interest.

[0097] At the very least, to mitigate some of the limitations mentioned above and ensure that robust but vendor-specific ML models are trained, vendor-specific data should benefit from being artificially diversified and augmented on a vendor-by-vendor basis before being used to train vendor-based ML models. The process of artificially augmenting training data is called data augmentation, and the success of the procedure depends on two main factors: the quantity and quality of the initial training data, as well as the augmentation algorithm and its design assumptions. Nevertheless, there are no concrete proposals at this stage regarding how the required training data should be collected from different UEs within the network to enable vendor-independent ML-enabled solutions.

[0098] Some embodiments of the examples provide a technique in which vendor-specific training data (hereinafter referred to as domain-variable data) is diversified by using data from other vendors without exposing / sharing domain-variable datasets between vendors. For this purpose, the domain-variable data is first depended upon, i.e., deprived of its vendor-specific properties. Three types of NR elements are included, and the coupling of functions can be carried out by a single NR element.

[0099] ML Coordinator Function (NR-C) - An NR network element whose role is to aggregate training data collected by different UEs and / or from multiple UE vendors, and to define the format of the vendor-independent or converted training data that each UE should transfer back to the NR-C. The NR-C may also be gNB-CU, NRT-RIC, NWDAF, etc.

[0100] ML Data Collector Function (NR-E) - An NR network element whose role is to collect / modify raw data in a first or vendor-specific format as instructed by NR-C, and to transfer the data to NR-T or NR-C using a vendor-independent or converted format. NR-E may also be a UE, gNB-CU, RT-RIC, RSU, etc. NR-Ek represents an NR-E that collects vendor-specific training data for k.

[0101] ML Training Function (NR-T) - An NR network element that combines training data from different sources, such as multiple NR-Es, to train vendor-specific ML functions. An NR-T may be an NWDAF, LMF, serving gNB, or UE. An NR-T may reside within the same network element as its corresponding NR-E, such as a gNB or UE. NR-Tm represents an NR-T that trains an ML model for vendor m.

[0102] Provide training data from Vendor 2 to Vendor 1. An example embodiment is shown in Figure 2, where a coordinator function configures data collector functions to provide training data to the training function. One of the data collector functions is from the same vendor as the training function and can therefore provide its training data in the format expected by this training function. Another data collector function is from a different vendor and is therefore instructed to collect specified training data, convert this training data to a specified format, and provide this converted training data to the training function. The training function then converts the converted training data to match the format of the training data provided by the data collector function from the same vendor as the training function, combines the training data, and uses this combined training data to train an ML model.

[0103] NR-C constructs elements NR-Ek, k=1..K, for each vendor k, so that it provides its training data in a given format.

[0104] Therefore, in step S10, NR_C instructs NR_E1 to provide NR_T1 with training data in a vendor-specific format called domain variable data (DVD), when the training data and ML training are for the same vendor. In step S20, NR_E1 collects the training data DVD1 in vendor-specific format, and in step S30, reports this training data DVD1 to NT_T1.

[0105] In step S40, NR_C instructs NR_E2 to provide training data to NR_T2 in a vendor-independent format called Domain Invariant Data (DID) format if the vendor requiring training is different from the vendor from which the data is collected. NR_C defines how the DID is obtained on each NR-Ek side by providing vendor-versus-invariant conversion (VIC) or transformation details, as described in more detail below.

[0106] In step S50, each NR-Ek, k=1.., K, converts domain variable data k(DVD(k)) to DID using the VIC provided by NR_C. The DID has a unique format (format, size, type such as integer, real, etc.) and may be scenario-specific. For example, a DID for beamforming may be a 2D or 3D matrix of real numbers where each entry is the RSRP of RS at a given (time, frequency) or (time, frequency, space) position, and a DID for positioning enhancement may be a 3D matrix of complex numbers where each entry is the channel gain at a given (time, delay, space) position, generating a universally understood / agreeable DID format that can be used across different UE vendors / domains. DVD-to-DID conversion (VIC) is configured by NR-C and typically aims to deprive the DVD of sensitive information (UE-specific data payload, symbols, identifiers), as well as vendor-specific artifacts such as UE-specific TX / RX delays and beam offsets. Note that the exact form of the VIC conversion that will be applied to the DVD to obtain the DID is derived in NR_E, which performs the conversion. VIC parameterization may be configured entirely / as a whole by NR-C, and can limit the DID format, DID reporting cycle, and target VIC performance, with performance metrics depending on the use case.

[0107] In step S60, NR-Ek sends its DIDk to NR-Tm. Alternatively, the DID may be sent first to NR-C, which then forwards the DID to each NR-Tm.

[0108] In step S70, NR-Tm diversifies DVDm using a DID from domain k, k=1...K, where m≠k, and each DID(k), k=1...K, is used to reconstruct DID(k) into DVD(m). The reconstruction or conversion process consists of inputting DID(k) into a module that applies domain m-specific transformations to the DID and outputs an approximate DVD(m) called reconstructed DVD:R-DVD(k→m). The conversion from DID to R-DVD (Invariant vs. Variable Conversion (IVC)) is the inverse of VIC, and involves transforming the DID into DVD format and also including domain-specific information where available (depending on the use case). This conversion ensures that the R-DVD has the same properties and format as the DVDm of a particular UE vendor m. The exact format of the IVC conversion needs to be known only to the vendor-specific features (NR_T and / or NR_E).

[0109] In step S80, NR-Tm uses all R-DVD(k→m), k=1..K, and combines them with the original DVD(m) to form a superset C-DVD(m) = Combine{R-DVD(k→m), ∀k=1..K, DVD(m)}. The Combine function can consist of various operations such as superimposing, averaging, and filtering of datasets.

[0110] In step S90, C-DVD(m) is extended to obtain the final training dataset for domain m, and in step S100, a domain m-specific ML model is trained. Such extensions typically include concatenation of datasets, random mixing of datasets, and similar methods.

[0111] In other words, vendor 1 requests training of an ML module. NR_T1 is the function that trains a vendor 1-specific ML model. NR_T1 is configured by NR_C to collect DVD1 from NR_E1, where NR_E1 belongs to vendor 1—where both the training and data belong to the same vendor, so the data can be shared directly; and to collect DID2 from NR_E2, where NR_E2 belongs to vendor 2, and therefore the data needs to be vendor-dependent before it can be shared. NR_E1 and NR_E2 are configured by NR_C to collect training data and share the training data with NR_T1. NR_E2 is configured by NR_C to transform its own DVD2 into DID2 by applying its own VIC2. NR_T1 is constructed by NR_C to apply IVC1 to DID2, so that DID2 is converted to a reconstructed DVD, R-DVD(2→1), that is, data that vendor 1 can use for training and which originates from vendor 2's NR elements. NR_T1 then combines DVD1 and R-DVD(2→1), and such a combining function is generally represented as Combine1. The function Combine1 can perform any of the operations on DVD1 and R-DVD(2→1), such as superposition, averaging, filtering, pruning, puncturing, scaling, normalization, or a combination of the above operations.

[0112] An example embodiment is shown in Figure 3, where the coordinator function configures data collector functions to provide training data to the coordinator function. One of the data collector functions is from the same vendor as the training function and can therefore provide its training data in the format expected by the training function, but is instructed to collect specified training data, convert this training data to a specified format, and provide this converted training data to the coordinator function. Another data collector function is from a different vendor and is instructed to collect specified training data, convert this training data to a specified format, and provide this converted training data to the coordinator function. The coordinator function then combines the training data and provides it to the training function. The training function converts the converted training data to match the format of the training data provided by the data collector function from the same vendor as the training function, and uses this training data to train an ML model.

[0113] Specifically, the NR-C function collects DID(k), combines them into a combined DID(C-DID), which is then sent to the vendor's NR-T. NR_C configures the target DID format for all NR_E from which data will be collected. Each NR_E is assumed to be able to derive the corresponding VIC conversion, knowing the DID format and its own DVD format. NR_T knows the inverse IVC corresponding to the vendor from which the training data will be generated.

[0114] Therefore, NR_C instructs NR_E1 to provide NR_T1 with training data in a vendor-specific format called Domain Variable Data (DVD) when the training data and ML training are for the same vendor. In step S20, NR_E1 collects the training data DVD1 in the vendor-specific format, and in step S30, reports this training data DVD1 to NT_T1.

[0115] In steps S110 and S140, NR_C instructs NR_E1 and NR_E2 to provide training data to NR_C in a vendor-independent format called Domain Invariant Data (DID) format. NR_C defines how the DID is obtained on each NR-Ek side by providing vendor-versus-invariant conversion (VIC) or transformation details, as described in more detail below.

[0116] In steps S120 and S150, each NR-Ek, k=1.., K, converts domain variable data k(DVD(k)) to DID using the VIC provided by NR_C. The DID has a unique format (format, size, type such as integer, real, etc.) and may be scenario-specific. For example, a DID for beamforming may be a 2D or 3D matrix of real numbers where each entry is the RSRP of RS at a given (time, frequency) or (time, frequency, space) position, and a DID for positioning enhancement may be a 3D matrix of complex numbers where each entry is the channel gain at a given (time, delay, space) position, generating a universally understood / agreeable DID format that can be used across different UE vendors / domains. DVD-to-DID conversion (VIC) is configured by NR-C and typically aims to deprive the DVD of sensitive information (UE-specific data payload, symbols, identifiers), as well as vendor-specific artifacts such as UE-specific TX / RX delays and beam offsets. Note that the exact form of the VIC conversion that will be applied to the DVD to obtain the DID is derived in NR_E, which performs the conversion. VIC parameterization may be configured entirely / as a whole by NR-C, and can limit the DID format, DID reporting cycle, and target VIC performance, with performance metrics depending on the use case.

[0117] In steps S130 and S160, each NR-Ek sends its DIDk to NR-C.

[0118] In step S170, NR-C combines all DIDk values ​​to form a superset C-DID(m)=Combine{DID(k)}. The Combine function can consist of various operations such as overlaying datasets, averaging, and filtering.

[0119] In step S180, C-DID is reported to NR_T1.

[0120] In step S190, using the C-DID, NR-T1 diversifies the C-DID to reconstruct it into R-DVD1. The reconstruction or conversion process consists of inputting the C-DID into a module that applies domain-specific transformations to the DID and outputs an approximate DVD called the reconstructed DVD: R-DVD. The DID to R-DVD conversion (Invariant vs. Variable Conversion (IVC)) is the inverse of VIC and involves transforming the DID into DVD format and also including domain-specific information where available (depending on the use case). This conversion ensures that the R-DVD has the same properties and format as the DVDm of a particular UE vendor m. The exact format of the IVC conversion needs to be known only to the vendor-specific functions (NR_T and / or NR_E).

[0121] In step S200, R-DVD(m) is extended to obtain the final training dataset for domain m, and in step S210, the domain m-specific ML model is trained.

[0122] Provide training data from Vendor 2 to Vendor 1, and from Vendor 1 to Vendor 2. An example embodiment is shown in Figure 4, in which the data collector function is configured such that the coordinator function provides training data to the training functions of multiple vendors. This example embodiment can be combined with the example embodiments described above. The data collector function is instructed to collect specified training data, convert this training data to a specified format, and provide this converted training data to the training function. The training function then combines the training data to match the format of the vendor's training data and uses this training data to train its ML model.

[0123] Therefore, NR_C instructs NR_E1 and NR_E2 to provide training data to NR_T1 and NR_T2 in a vendor-specific format called Domain Variable Data (DVD). In step S20, NR_E1 collects training data DVD1 in the vendor-specific format, and in step S30, reports this training data DVD1 to NT_T1. In step S220, NR_E2 collects training data DVD2 in the vendor-specific format, and in step S230, reports this training data DVD1 to NT_T2.

[0124] In steps S240 and S280, NR_C instructs NR_E1 and NR_E2 to provide training data to NR_T1 and NR_T2 in a vendor-independent format called Domain Invariant Data (DID) format. NR_C defines how the DID is obtained on each NR-Ek side by providing vendor-versus-invariant conversion (VIC) or transformation details, as described in more detail below.

[0125] In steps S250 and S290, each NR-Ek, k=1.., K, converts domain variable data k(DVD(k)) to DID using the VIC provided by NR_C. The DID has a unique format (format, size, type such as integer, real, etc.) and may be scenario-specific. For example, a DID for beamforming may be a 2D or 3D matrix of real numbers where each entry is the RSRP of RS at a given (time, frequency) or (time, frequency, space) position, and a DID for positioning enhancement may be a 3D matrix of complex numbers where each entry is the channel gain at a given (time, delay, space) position, generating a universally understood / agreeable DID format that can be used across different UE vendors / domains. DVD-to-DID conversion (VIC) is configured by NR-C and typically aims to deprive the DVD of sensitive information (UE-specific data payload, symbols, identifiers), as well as vendor-specific artifacts such as UE-specific TX / RX delays and beam offsets. Note that the exact form of the VIC conversion that will be applied to the DVD to obtain the DID is derived in NR_E, which performs the conversion. VIC parameterization may be configured entirely / as a whole by NR-C, and can limit the DID format, DID reporting cycle, and target VIC performance, with performance metrics depending on the use case.

[0126] In steps S260, S270, S300, and S310, each NR-Ek sends its DIDk to both NR_T1 and NR_T2.

[0127] In steps S320 and S330, NR-T1 and NR_T2 combine all DIDk values ​​to form a superset C-DID(m)=Combine{DID(k)}. The Combine function can consist of various operations such as overlaying, averaging, and filtering of datasets.

[0128] In steps S340 and S350, C-DID(m) is used to train a domain m-specific ML model.

[0129] Alternatively, C-DID, NR-T1, and NR_T2 are used to diversify C-DID in order to reconstruct it into R-DVD1 and R-DVD1. The reconstruction or conversion process consists of inputting C-DID into a module that applies domain-specific transformations to the DID and outputs an approximate DVD called the reconstructed DVD: R-DVD. DID to R-DVD conversion (Invariant vs. Variable Conversion (IVC)) is the inverse of VIC and involves transforming the DID into DVD format and also including domain-specific information where available (depending on the use case). This conversion ensures that the R-DVD has the same properties and format as DVDm of a particular UE vendor m. The exact form of the IVC conversion needs to be known only to vendor-specific features (NR_T and / or NR_E). The R-DVD(m) is then extended to obtain the final training dataset for domain m and to train a domain m-specific ML model.

[0130] Vendor Y provides training data to Vendor A. An example embodiment is shown in Figure 5, in which a coordinator function from one vendor is configured to provide training data to a training function of another vendor, thereby configuring a data collector function from another vendor. This example embodiment can be combined with the example embodiments described above. The data collector function is instructed to collect specified training data, convert this training data to a specified format, and provide this converted training data to the training function. The training function then combines the training data to match the format of this vendor's training data and uses this training data to train its ML model.

[0131] In step S20, NR_EY collects training data DVDY in a vendor-specific format.

[0132] In step S370, NR_C instructs NR_EY to provide training data to NR_TA in a vendor-independent format called Domain Invariant Data (DID) format. NR_C defines how the DID is obtained on each NR-Ek side by providing vendor-versus-invariant conversion (VIC) or transformation details as described above.

[0133] In step S380, NR_C reports a DID to R-DVD conversion (Invariant vs. Variable Conversion (IVC)) including domain-specific information, if available (depending on the use case). This conversion ensures that the R-DVD has the same properties and format as DVDm from a specific UE vendor m. In this example, IVC_A provides a DID to R-DVD_A conversion.

[0134] In step S390, each NR-Ek, k=1.., K, converts domain variable data k(DVD(k)) to DID using the VIC provided by NR_C. The DID has a unique format (format, size, type such as integer, real, etc.) and may be scenario-specific. For example, a DID for beamforming may be a 2D or 3D matrix of real numbers where each entry is the RSRP of RS at a given (time, frequency) or (time, frequency, space) position, and a DID for positioning enhancement may be a 3D matrix of complex numbers where each entry is the channel gain at a given (time, delay, space) position, generating a universally understood / agreeable DID format that can be used across different UE vendors / domains. DVD-to-DID conversion (VIC) is configured by NR-C and typically aims to deprive the DVD of sensitive information (UE-specific data payload, symbols, identifiers), as well as vendor-specific artifacts such as UE-specific TX / RX delays and beam offsets. Note that the exact form of the VIC conversion that will be applied to the DVD to obtain the DID is derived in NR_E, which performs the conversion. VIC parameterization may be configured entirely / as a whole by NR-C, and can limit the DID format, DID reporting cycle, and target VIC performance, with performance metrics depending on the use case.

[0135] In step S400, each NR-Ek sends its DIDk to NR_TA.

[0136] In step S410, using the DID, NR-TA diversifies the DID to reconstruct it into R-DVD_A. The reconstruction or conversion process consists of inputting the DID into a module that applies domain-specific transformations to the DID and outputs an approximate DVD called the reconstructed DVD: R-DVD. This conversion ensures that the R-DVD has the same properties and format as DVDm from a specific UE vendor m. The R-DVD_A is then expanded to obtain the final training dataset for domain m, and in step S400, a domain m-specific ML model is trained.

[0137] Sidelink (SL) positioning As illustrated in Figure 6, in one example embodiment, for SL positioning, Vendor 1 UE trains an ML-based position estimator using its own data DVD1 and DID2 collected from Vendor 2 UE. While this example concerns positioning, it will be understood that this technique is applicable to other use cases.

[0138] In step S430, vendor 1 UE collects time-frequency measurements of DL PRS and stores them in matrix DVD1.

[0139] In step S440, Vendor 2 UE is instructed by Vendor 1 UE to collect DL PRS time-frequency measurements and store them in matrix DVD2, where, DVD2(i,j) = channel frequency response at frequency i*Fs2 and time j*Ts2, where Fs2 and Ts2 are both sampling frequencies and times specific to vendor 2UE, i≠a and j≠b.

[0140] Using the new SL PSSCH IE, Vendor 1 UE configures Vendor 2 UE to report DID2, and DID2 is generated by the VIC2 converter for all Vendor 2 UEs.

[0141] Vendor 1 UE configures the VIC2 parameters. For example, VIC2 should output DID2, where, DID2(i,j) = i*Fs1 and CFR at time i*Ts1, in other words, DVD2 should be resampled at rate Fs1 and its resolution, with Ts2 changed to Ts1. The RX beam response of Vendor 2 UE, i.e., W2, is removed from DVD2. For example, VIC2 should apply the DVD2 conversion as follows: DID2 = inv(W2) * DVD2.

[0142] In step S450, Vendor 2 UE collects DVD2 and applies VIC2 as instructed, and in step S460, reports DID2 to LMF.

[0143] In step S470, Vendor 1 UE reconstructs the R-DVD (2→1) using DID2 and converter IVC2. In other words, Vendor 1 UE applies its own response to DID2 to artificially generate how the Vendor 1 UE data should appear in resource index (i,j).

[0144] Next, in step 480, Vendor 1 UE combines DVD1 with R-DVD(2→1) to create C-DVD(1). For example, Vendor 1 UE can superimpose two matrices or calculate two average responses.

[0145] In step S490, Vendor 1 UE produces an expanded DVD1 using C-DVD1 and a preferred modern expansion method (scaling, translation, etc.).

[0146] In step S500, the expanded set is then used to train a preferred modern ML-based localization estimator (e.g., a deep neural network (DNN) with a normalized linear unit (ReLU) activation function).

[0147] UE-assisted DL positioning As illustrated in Figure 7, one embodiment of an example includes UE-assisted DL positioning, where NR_T (or associated Network Data Analysis (NWDAF) function) trains an ML-based position estimator for UE vendor 1 using the DVD of vendor 1's UE and DID2 collected from vendor 2 using the techniques described in Figure 3.

[0148] Vendor 1 UE is instructed by NR-C to collect time-frequency measurements of the DL PRS and store them in matrix DVD1, where, DVD1(a,b) = channel frequency response at frequency a*Fs1 and time b*Ts1, where Fs1 and Ts1 are sampling frequency and time, respectively, both specific to vendor 1 UE.

[0149] A new LPP IE is used for Vendor 1 UE to report the DVD to NR_T.

[0150] Vendor 2 UE is instructed by NR_C to collect DL PRS time-frequency measurements and store them in matrix DVD2, where, DVD2(i,j) = channel frequency response at frequency i*Fs2 and time j*Ts2, where Fs2 and Ts2 are the sampling frequency and time, respectively, which are both specific to vendor 2UE, and i≠a and j≠b.

[0151] A new LPP IE is used, and therefore NR_C configures the Vendor 2 UE to report DID2, where DID2 is generated by the converter VIC2 for all Vendor 2 UEs.

[0152] NR_C constitutes the VIC2 parameters. For example, VIC2 should output DID2, where, DID2(i,j) = i*Fs1 and CFR at time i*Ts1, in other words, DVD2 should be resampled at rate Fs1 and its resolution, with Ts2 changed to Ts1. The RX beam response of Vendor 2 UE, i.e., W2, is removed from DVD2. For example, VIC2 should apply the DVD2 conversion as follows: DID2 = inv(W2) * DVD2.

[0153] Vendor 2 UE collects DVD2, applies VIC2 as instructed, and reports DID2 to NR_C.

[0154] NR_T reconstructs the R-DVD (2→1) using DID2 and the converter IVC2. In other words, NR_T applies a vendor-specific response to DID2 to artificially generate how the vendor-specific data should appear in resource index (i,j).

[0155] NR_T combines DVD1 with R-DVD(2→1) to create C-DVD(1). For example, NR_T can superimpose two matrices or calculate the average response of two matrices.

[0156] NR_T produces an extended DVD1 using C-DVD1 and preferred modern extension methods (scaling, translation, etc.).

[0157] The expanded set is then used to train a preferred modern ML-based location estimator (e.g., a DNN with a ReLU activation function).

[0158] Those skilled in the art will readily recognize that various steps of the above-described methods can be carried out by a programmed computer. In this specification, some embodiments are also intended to cover program storage devices, such as digital data storage media, which are machine- or computer-readable and encode machine-executable or computer-executable programs of instructions, the instructions carrying out some or all of the steps of the above-described methods. The program storage devices may be, for example, digital memory, magnetic storage media such as magnetic disks and magnetic tapes, hard drives, or optically readable digital data storage media. Embodiments are also intended to cover computers programmed to carry out the steps of the above-described methods. The term "non-temporary," as used herein, refers to the limitations of the medium itself (i.e., tangible, non-signaling) as opposed to the limitations of data storage (e.g., RAM vs. ROM).

[0159] As used in this application, the term "circuit equipment" means, (a) Hardware-only circuit embodiments (such as embodiments in analog and / or digital circuit devices only), (b) (where applicable), (i) combinations of analog and / or digital hardware circuits and software / firmware, (ii) Any part of a hardware processor with software (including a digital signal processor, software, and memory that work together to enable a device such as a mobile phone or server to perform various functions) such combinations of hardware circuits and software, (c) Hardware circuits and / or processors, such as a microprocessor or a part of a microprocessor, that require software (e.g., firmware) for operation, but may not be present when the software is not required for operation. It represents one, more, or all of the following.

[0160] This definition of circuit equipment applies to all uses of the term in this application, including those included in any claim. For further example, as used in this application, the term circuit equipment also covers embodiments of hardware circuits or processors (or a number of processors) alone, or parts of hardware circuits or processors, and their accompanying software and / or firmware. The term circuit equipment also covers, for example, and, where applicable to a particular claim, baseband integrated circuits or processor integrated circuits for mobile devices, or similar integrated circuits in servers, cellular network devices, or other computing or network devices.

[0161] While exemplary embodiments of the present invention have been described in the preceding paragraph with reference to various examples, it should be understood that modifications to the examples shown may be made without departing from the claimed scope of the invention.

[0162] The features described in the previous explanation may be used in combinations other than those explicitly described.

[0163] While features are described by referring to specific characteristics, these features may be achievable through other characteristics, regardless of whether they are described or not.

[0164] While the features are described with reference to specific embodiments, these features may also be present in other embodiments, whether or not they are described.

[0165] While efforts have been made in the foregoing specification to draw attention to these features of the present invention that are deemed to be of particular importance, it should be understood that the applicant seeks protection for any patentable features or combinations of features mentioned herein and / or shown in the drawings, whether or not they are particularly emphasized.

Claims

1. A coordinator function configured to send a first configuration message to a first collector function, The first configuration message includes information for configuring the first collector function to collect first training data, convert the first training data in the first collector function to generate converted first training data, and report the converted first training data. The aforementioned information is, Information specifying the data format of the converted first training data, including information specifying at least one of the sampling resolution, array shape, and length, Information specifying the reporting cycle for the converted first training data, From the first training data described above, parameters relating to vendor-to-invariant conversion for removing specified vendor-specific data or specified vendor-specific artifacts are obtained. Includes, The first configuration message causes the first collector function to remove the specified vendor-specific data or specified vendor-specific artifacts from the first training data based on the vendor-versus-invariant conversion parameters. Coordinator function A device equipped with the following features.

2. The first configuration message is, By removing specified vendor-specific data or artifacts, the first training data is transformed to generate the transformed first training data, or Report the converted first training data in the converted training data format or at a specified reporting interval. The apparatus according to claim 1, including information for configuring the first collector function to perform the following:

3. The apparatus according to claim 1, wherein the first configuration message includes information for configuring the first collector function to report the converted first training data to at least one of the coordinator function, training function, first training function, and second training function.

4. The coordinator function is configured to send a second configuration message to a second collector function. The second configuration message includes information for configuring the second collector function to collect second training data, convert the second training data in the second collector function to generate converted second training data, and report the converted second training data. The information contained in the second configuration message is: Information specifying the data format of the converted second training data, Information specifying the reporting cycle for the converted second training data, From the second training data described above, parameters for vendor-to-invariant conversion to remove specified vendor-specific data or specified vendor-specific artifacts are obtained. The apparatus according to claim 1, including the following:

5. The apparatus according to claim 4, wherein the coordinator function is configured to combine the converted first training data received from the first collector function and the converted second training data received from the second collector function to form combined converted training data, and to send the combined converted training data to the training function.

6. The coordinator function is configured to send configuration messages to a collector function provided by a vendor common to both the coordinator function and the first collector function. The apparatus according to claim 1, wherein the configuration message includes an instruction for configuring the collector function to collect training data and report the training data.

7. The aforementioned coordinator function is configured to send conversion configuration messages to the training function. The conversion configuration message includes information for configuring the training function to convert the received converted training data or the received combined converted training data into converted training data. The information included in the conversion configuration message includes parameters relating to invariant versus variable conversions, The apparatus according to claim 1, wherein the invariant-to-variable conversion is for converting the converted training data or the combined converted training data into training data in a vendor-specific format corresponding to the training function.

8. The apparatus according to claim 7, wherein the conversion configuration message includes information for configuring the training function to convert the received transformed training data or the received combined transformed training data into the converted training data by performing at least one of the following operations: expansion, averaging, filtering, pruning, puncturing, normalization, scaling, translation, and projection.

9. The aforementioned coordinator function, Sending a first conversion configuration message to a first training function, wherein the first conversion configuration message includes information for configuring the first training function to convert the received converted training data or received combined converted training data into first converted training data. Sending a second conversion configuration message to a second training function, wherein the second conversion configuration message includes information for configuring the second training function to convert the received converted training data or received combined converted training data into second converted training data. It is configured to do the following: The information contained in the first conversion configuration message and the second conversion configuration message includes parameters relating to invariant versus variable conversions. The apparatus according to claim 1.

10. At least one of the first configuration message and the converted training data is transmitted via at least one of the following: physical sidelink shared channel, physical downlink shared channel, physical uplink shared channel, F1 interface, Xn interface, NG interface, O-RAN A1 interface, O-RAN E1 interface, O-RAN E2 interface, and O-RAN F1 interface. The apparatus according to any one of claims 1 to 9.

11. A step of sending a first configuration message to a first collector function, The first configuration message includes information for configuring the first collector function to collect first training data, convert the first training data in the first collector function to generate converted first training data, and report the converted first training data. The aforementioned information is, Information specifying the data format of the converted first training data, including information specifying at least one of the sampling resolution, array shape, and length, Information specifying the reporting cycle for the converted first training data, From the first training data described above, parameters relating to vendor-to-invariant conversion for removing specified vendor-specific data or specified vendor-specific artifacts are obtained. Includes, The first configuration message causes the first collector function to remove the specified vendor-specific data or specified vendor-specific artifacts from the first training data based on the vendor-versus-invariant conversion parameters. Methods that include...

12. A collector function configured to receive configuration messages from a coordinator function, The configuration message includes information for configuring the collector function to collect training data, convert the training data in the collector function to generate converted training data, and report the converted training data. The aforementioned information is, Information specifying the data format of the converted training data, including information specifying at least one of the sampling resolution, array shape, and length, Information specifying the reporting cycle for the converted training data, From the aforementioned training data, parameters relating to vendor-to-invariant conversion for removing specified vendor-specific data or specified vendor-specific artifacts are obtained. Includes, The collector function is configured to generate the converted training data by removing the specified vendor-specific data or specified vendor-specific artifacts from the training data based on the parameters relating to the vendor-versus-invariant conversion. Collector function A device equipped with the following features.

13. This is the step of receiving a configuration message from the coordinator function. The configuration message includes information for configuring the collector function to collect training data, convert the training data in the collector function to generate converted training data, and report the converted training data. The aforementioned information is, Information specifying the data format of the converted training data, including information specifying at least one of the sampling resolution, array shape, and length, Information specifying the reporting cycle for the converted training data, From the aforementioned training data, parameters relating to vendor-to-invariant conversion for removing specified vendor-specific data or specified vendor-specific artifacts are obtained. Steps including, The steps include generating the converted training data by removing the specified vendor-specific data or specified vendor-specific artifacts from the training data based on the vendor-versus-invariant conversion parameters, The steps include reporting the converted training data and Methods that include...

14. A training function configured to receive conversion configuration messages from the coordinator function, The conversion configuration message includes information for configuring the training function to convert the received converted training data or the received combined converted training data into converted training data. The aforementioned information includes parameters relating to invariant versus variable conversions, The aforementioned invariant-to-variable conversion is for converting the converted training data or the combined converted training data into training data in a vendor-specific format corresponding to the training function. Training function A device equipped with the following features.

15. This is the step where you receive a conversion configuration message from the coordinator function. The aforementioned conversion configuration message includes information for configuring the training function to convert the received converted training data or the received combined converted training data into converted training data. The aforementioned information includes parameters relating to invariant versus variable conversions, The invariant-to-variable conversion is for converting the converted training data or the combined converted training data into training data in a vendor-specific format corresponding to the training function, step Methods that include...

Citation Information

Patent Citations

  • Apparatus and method for man-machine interaction of AI / ML training hosts

    CN114443556A

  • Functional architecture and interface for non-real-time ran intelligent controller

    US20210337420A1

  • Wireless signaling in federated learning for machine learning components

    US20220182802A1

  • Distributed Ledger for Additive Manufacturing in Value Chain Networks

    US20220197247A1