B-type model identification procedure for model-id based lcm

Abstract

Systems and methods can use a B-type model identification process for model ID based lifecycle management (LCM). Some embodiments can include training a new model for wireless communication. A user equipment (UE) or network node can transmit an indication including a new model identification (ID) associated with the new model. The UE or network node can also transmit the new model with the new model ID.

Description

Technical Field

[0001] This application relates in its entirety to wireless communication systems, including model identifiers via over-the-air signaling. Background Technology

[0002] Wireless mobile communication technologies use various standards and protocols to transmit data between base stations and wireless communication devices. For example, wireless communication system standards and protocols may include, for instance, 3GPP Long Term Evolution (LTE) (e.g., 4G), 3GPP New Radio (NR) (e.g., 5G), and the Institute of Electrical and Electronics Engineers (IEEE) 802.11 standard for Wireless Local Area Networks (WLANs) (often referred to as Wi-Fi within the industry organization). ® ).

[0003] As envisioned by 3GPP, different wireless communication system standards and protocols can use various radio access networks (RANs) for communication between RAN base stations (sometimes referred to as RAN nodes, network nodes, or simply nodes) and wireless communication equipment called user equipment (UEs). 3GPP RANs can include, for example, Global System for Mobile Communications (GSM), Enhanced Data Rate GSM Evolution (EDGE) RAN (GERAN), Universal Terrestrial Radio Access Network (UTRAN), Evolved Universal Terrestrial Radio Access Network (E-UTRAN), and / or Next Generation Radio Access Network (NG-RAN).

[0004] Each RAN can use one or more Radio Access Technologies (RATs) to perform communication between the base station and the UE. For example, GERAN implements the GSM and / or EDGE RAT, UTRAN implements the Universal Mobile Telecommunications System (UMTS) RAT or other 3GPP RATs, E-UTRAN implements the LTE RAT (sometimes simply referred to as LTE), and NG-RAN implements the NR RAT (this NR RAT is sometimes referred to herein as the 5G RAT, 5G NR RAT, or simply NR). In some deployments, E-UTRAN may also implement the NR RAT. In some deployments, NG-RAN may also implement the LTE RAT.

[0005] The base stations used by a RAN can correspond to that RAN. An example of an E-UTRAN base station is an Evolved Universal Terrestrial Radio Access Network (E-UTRAN) Node B (also commonly referred to as Evolved Node B, Enhanced Node B, eNodeB, or eNB). An example of an NG-RAN base station is a Next Generation Node B (sometimes also called gNode B or gNB).

[0006] The RAN provides communication services to external entities through its connection with the core network (CN). For example, E-UTRAN can utilize the evolved packet core (EPC), while NG-RAN can utilize the 5G core network (5GC). Attached Figure Description

[0007] To facilitate the identification of any particular element or action in the discussion, one or more of the most significant digits in the figure reference numerals refer to the figure number in which the element was first introduced.

[0008] Figure 1 A table illustrating the delivery / transmission scenarios based on some implementation schemes is provided.

[0009] Figure 2 An example signal flow diagram of model transmission for type B1 with a known model structure is illustrated according to some implementation schemes.

[0010] Figure 3 An example signal flow diagram of model transmission for type B2 with a known model structure is illustrated according to some implementation schemes.

[0011] Figure 4 An example signal flow diagram for model transmission with an unknown model structure for type B1 is illustrated according to some implementation schemes.

[0012] Figure 5 An example signal flow diagram for model transmission with an unknown model structure for type B2 is illustrated according to some implementation schemes.

[0013] Figure 6 Example signal flow diagrams for model updates for type B1 are illustrated according to some implementation schemes.

[0014] Figure 7 Example signal flow diagrams for model updates for type B2 are shown according to some implementation schemes.

[0015] Figure 8 Examples of sub-models for combining fine-tuning model transfers are shown according to some implementation schemes.

[0016] Figure 9 A signal flow diagram illustrating model identification for additional network-side conditions is shown according to some implementation schemes.

[0017] Figure 10 A signal flow diagram illustrating model identification for additional conditions on the UE side is shown according to some implementation schemes.

[0018] Figure 11 A method for a UE according to an embodiment of this document is illustrated.

[0019] Figure 12 A method for network nodes according to the implementation scheme of this paper is illustrated.

[0020] Figure 13 An example architecture of a wireless communication system according to the implementation scheme disclosed herein is illustrated.

[0021] Figure 14 A system for performing signaling between a wireless device and a network device according to an embodiment disclosed herein is illustrated. Detailed Implementation

[0022] Various implementations are described with respect to the UE. However, references to the UE are provided for illustrative purposes only. The example implementations can be used with any electronic component capable of establishing a connection to a network and configured with hardware, software, and / or firmware for exchanging information and data with the network. Therefore, the UE as described herein is used to represent any suitable electronic component.

[0023] The goals of wireless communication systems include providing reliable, efficient, and secure communication between the user equipment (UE) and network nodes. Artificial intelligence (AI) / machine learning (ML) can be used to help achieve these goals. For example, AI technologies can be used within wireless systems in several ways, including network optimization. AI can be used to optimize network performance by analyzing data, predicting future traffic patterns, and identifying congested areas. This can help improve network efficiency and reduce downtime. Furthermore, AI can be used to automate network operations such as provisioning, configuration, and optimization. This can help reduce costs and improve operational efficiency.

[0024] Lifecycle Management (LCM) refers to the use of AI technology to automate and optimize various aspects of the entire lifecycle of network components, services, or applications. The lifecycle of a network component typically includes phases such as planning, deployment, operation, and maintenance. LCM leverages AI capabilities to simplify and automate tasks associated with these phases, ensuring efficient network management and optimization. With AI-driven LCM, network operators can benefit from advanced analytics, machine learning algorithms, and automation to plan and optimize network resources. AI algorithms can analyze network performance data, predict traffic patterns, and optimize resource allocation to improve efficiency and quality of service. LCM can also be used to automate configuration and deployment. For example, AI can automate the configuration and deployment of network elements, reducing human intervention and minimizing human error. Furthermore, AI-enabled monitoring and diagnostic tools can proactively detect anomalies and potential problems, enabling rapid problem resolution and reducing network downtime. Additionally, by analyzing network data, AI algorithms can predict potential failures or performance degradation in network components, allowing operators to proactively schedule maintenance activities, optimize performance, and reduce costs. Therefore, LCM can leverage artificial intelligence to automate and optimize network management tasks throughout the entire lifecycle of network components, thereby achieving improved efficiency, enhanced performance, and a better user experience.

[0025] Devices can utilize multiple AI / ML models for different conditions through a process called model selection or model switching. This approach allows devices to dynamically select the most suitable AI model based on the prevailing condition or requirements. These models can be used to enhance different aspects of wireless communication in various scenarios. The embodiments described in this paper provide apparatus, systems, and methods for transmitting and updating models based on model identifiers (IDs).

[0026] In some implementations, LCM can be based on a model ID. Such a model ID-based LCM relies on an accurate model identifier. Some implementations described herein provide a model identification process for model ID-based LCM. For model identification of the UE portion of a UE-side (single-side model) or dual-side model, implementations can categorize model identifier types as follows. Type A can be the case where the model is identified to the network (NW) (if applicable) and the UE (if applicable) without over-the-air signaling. The model can be assigned a model ID during model identification, which can then be referenced or used in over-the-air signaling after model identification.

[0027] Type B can refer to the case where the model is identified via air signaling. Two different types of Type B models can exist. Type B1 can refer to a model identification initiated by the UE, where the NW assists with the remaining steps of the model identification (if any). During model identification, the model can be assigned a model ID. Type B2 can refer to a model identification initiated by the NW, where the UE responds to the remaining steps of the model identification (if any), if applicable. Similar to Type B1, in Type B2, the model can be assigned a model ID during model identification.

[0028] In some implementations, when a model with a known structure at the UE is transmitted from the NW (e.g., case z4), the new model being identified (e.g., via type B2) may have the same structure as a previously identified model at the network and UE. In some implementations, the model ID may or may not be globally unique, and different types of model IDs may be created for a single model for various LCM purposes.

[0029] Figure 1 Table 102 illustrates the model delivery / transmission scenarios. The implementation scheme described herein is applicable to at least the following scenarios: model delivery / transmission targets to the UE for both the UE-side model and the UE portion of the dual-side model, training locations, and combinations of model delivery / transmission formats.

[0030] For UE-side model inference, to ensure consistency between training and inference regarding network-side additional conditions (if identified), the following options can be considered as potential methods (where feasible and necessary): alignment between the NW and UE sides regarding NW-side additional conditions via model identification; model training at the NW and transmission to the UE, where the model has already been trained under the additional conditions; providing the UE with information and / or indications regarding NW-side additional conditions; and consistency aided by monitoring (by the UE and / or NW monitoring the performance of UE-side candidate models / functions to select a model / function). Other methods are not excluded. This may not preclude the possibility that different methods can achieve the same functionality.

[0031] Some implementations described herein focus on Type B model identification, which is an over-the-air signaling procedure. Type A model identification is an offline method. In some implementations, Type 2 model identification can be used for model transfer. Some solutions provide model identification with a known model structure (model update). Some solutions provide model identification with an unknown model structure (new model). In some implementations, Type 2 model identification can be used for bilateral / unilateral model updates. Some solutions provide Type B1 for NW-side model identification. Some solutions provide Type B2 for UE-side model identification. In some implementations, Type B model identification can be used to achieve alignment between NW-side conditions and UE-side conditions.

[0032] Figure 2 Example signaling flowchart 202 illustrates a B1-type model transmission with a known model structure according to some implementation schemes. The illustrated implementation scheme is a B1-type model transmission initiated by the network (e.g., network node 204). The UE capability report may instruct the NW that UE 206 supports model parameter updates with a known model structure. If UE 206 does not indicate its support for model parameter updates in the UE capability report, the NW can avoid transmitting the model.

[0033] The NW can train a new model 220. Network node 204 can signal the new model ID 208 and / or metadata via Radio Resource Control (RRC). RRC signaling can be cell-specific signaling (e.g., System Information Block (SIB)) or UE-specific signaling. When transmitting the new model ID 208 and / or metadata via UE-specific signaling, the RRC message can be transmitted based on the UE capability report.

[0034] The model ID can be indicated in various ways based on the model ID design. In some implementations, if the model ID is defined by a version number, a new model can be signaled solely by the version ID. The other parts of the model ID can be the same to indicate that the model is a known model. The model structure can be implied by the previously identified model ID itself. In some implementations, the version number can be associated with a timestamp. In some implementations, if the model ID is not defined by a version number, the new model ID can be associated with a previously identified model ID in the signaling. The previously identified model ID can implicitly indicate a known model structure. In some implementations, a value tag can be added to the previously identified model ID. The value tag can indicate that it is an updated version with the same model structure.

[0035] UE 206 can check whether 210 supports the new model ID. Meta-information can provide more information about the updated model so that the UE can check its support and know the model structure. For example, meta-information can describe what the model is used for (e.g., the model is used for CSI compression). Meta-information can also describe the use case. For example, meta-information can describe whether the model is used indoors, outdoors, and / or for a specific antenna configuration. The described meta-information is provided as an example; other information about the model can be included in the meta-information.

[0036] If UE 206 is capable of supporting model parameter updates and the indicated known model ID is already supported at the UE, then UE 206 can send a request 212 to network node 204 to download the model. UE 206 can send request 212 via a Media Access Control Element (MAC CE) or an RRC message.

[0037] Network node 204 can transmit model file 214 along with a new model ID to UE 206. UE 206 may need to compile the downloaded model or request a UE server to compile the model. Compilation can occur at any implementation and may take some time to complete. When UE 206 is ready to run inference (e.g., after model compilation), UE 206 can transmit uplink message 216 to network node 204, indicating that the model is ready to run and UE 206 is ready to perform inference. Uplink message 216 may be UE Auxiliary Information (UAI), an uplink MAC CE for a model activation request, or UE capability signaling. Network node 204 can transmit activation / deactivation message 218 to UE 206 to activate or deactivate the model. UE 206 and network node 204 can continue with the remaining lifecycle management process (e.g., activating, monitoring, deactivating, or switching models).

[0038] Figure 3 Example signal flow diagram 302 illustrates a B2-type model transmission with a known model structure according to some implementation schemes. The illustrated implementation scheme is a B2-type model transmission initiated by UE 306. Offline training on the UE side can generate a new model 308 with updated model parameters. For example, UE 306 can use newly collected data to update the model parameters.

[0039] UE 306 may indicate the new model ID 310 and / or metadata to network node 304. The UE may transmit the indication via MAC CE or RRC messages. In some implementations, the indication of the new model ID 310 and / or metadata may be transmitted via UE capability reports.

[0040] The model ID can be indicated in various ways based on the model ID design. In some implementations, if the model ID is defined by a version number, a new model can be signaled solely by the version ID. The other parts of the model ID can be the same to indicate that the model is a known model. The model structure can be implied by the previously identified model ID itself. In some implementations, the version number can be associated with a timestamp. In some implementations, if the model ID is not defined by a version number, the new model ID can be associated with a previously identified model ID in the signaling. The previously identified model ID can implicitly indicate a known model structure. In some implementations, a value tag can be added to the previously identified model ID. The value tag can indicate that it is an updated version with the same model structure.

[0041] Network node 304 can check whether 312 supports the new model ID. Meta-information can provide more information about the updated model so that network node 304 can check its support and know the model structure. For example, meta-information can describe what the model is used for (e.g., the model is used for CSI compression). Meta-information can also describe the use case. For example, meta-information can describe whether the model is used indoors, outdoors, and / or for a specific antenna configuration. The described meta-information is provided as an example; other information about the model can be included in the meta-information.

[0042] If network node 304 supports model parameter updates, it can transmit uplink permission (e.g., request 314) for model transmission. UE 304 can transmit model file 316 along with the new model ID to network node 304 in the Physical Uplink Shared Channel (PUSCH). Uplink (UL) DCI, MAC CE, or RRC messages can be used to trigger UL model transmission.

[0043] If UE 206 is capable of supporting model parameter updates and the indicated known model ID is already supported at the UE, UE 206 can send a request 212 to network node 204 to download the model. UE 206 can send request 212 via a MAC CE or RRC message. When network node 304 is ready to perform inference, network node 304 can indicate the activation 318 of the model ID. UE 306 and network node 304 can continue the remaining lifecycle management process (e.g., activating, monitoring, deactivating, or switching models).

[0044] Figure 4 Example signaling flow diagram 402 illustrates a B1-type model transmission with an unknown model structure according to some implementation schemes. The illustrated implementation scheme is a B1-type model transmission initiated by the network (e.g., network node 404). The UE capability report can instruct the NW that UE 406 supports z5-type model transmission (e.g., model transmission with an unknown model).

[0045] If UE 406 has the capability to support z5 model transmission, the NW can signal the new model ID via RRC (e.g., New Model ID Announcement 408). Linking to the previous model ID may not be necessary. UE 406 can check 410 to see if the new model ID is supported.

[0046] UE 406 may send a request 412 to network node 404 to download the model. Request 412 may be sent via MAC CE or RRC message. Network node 404 may send model file 414 along with the new model ID to UE 406.

[0047] UE 406 may need to compile the downloaded model or request the UE server to compile the model. Compilation can occur at specific implementations and may take some time to complete. When UE 406 is ready to run inference (e.g., after model compilation), UE 406 may transmit an uplink message 416 to network node 404, indicating that the model is ready to run and UE 406 is ready to perform inference. Uplink message 416 may be a UAI, a UL MAC CE for model activation requests, or UE capability signaling. Network node 404 may transmit an activation / deactivation message 418 to UE 406 to activate or deactivate the model. UE 406 and network node 404 may continue with the remaining lifecycle management processes (e.g., activating, monitoring, deactivating, or switching models).

[0048] Figure 5 Example signal flow diagram 502 illustrates a B2-type model transmission with an unknown model structure according to some implementation schemes. The illustrated implementation scheme is a B2-type model transmission initiated by UE 506. A new model 508 can be generated by offline training on the UE or UE side.

[0049] UE 506 can indicate the availability of the new 510 model to network node 504. The UE can transmit the indication via MAC CE or RRC messages. The UE can indicate that the transmission is a z5 model transmission without a previously identified structure.

[0050] Network node 504 can check whether 516 NW supports the unknown structure (z5). If the unknown model structure is not supported, network node 504 can terminate the process. If the unknown model structure is supported, network node 504 can transmit UL approval for model transfer (e.g., request UL model transfer 512). UE 506 can transmit the model file along with the new model ID in the PUSCH (e.g., UL model transfer 514). In some implementations, the model file with the new model ID can be transmitted in the RRC message. Some implementations use fragmentation to adapt the new model file and new model ID into the RRC message.

[0051] When network node 504 is ready to perform inference, network node 504 can instruct the activation 518 of the model ID. UE 506 and network node 504 can continue the remaining lifecycle management process (e.g., activation, monitoring, deactivation, or switching of the model).

[0052] Figure 2 and Figure 3 It relates to the transmission of known models, and Figure 4 and Figure 5This relates to the transmission of unknown models. Some differences between these methods are the UE capabilities supporting z5 (unknown model) and whether the previously known model ID is linked to the new model ID. In some implementations, the two solutions can be combined into a common process.

[0053] Figures 2 to 5 The model transfer process is illustrated. Model transfer can be used to transfer new models and to provide updates to existing models. Some implementations can use... Figure 6 and Figure 7 The process in the middle provides model updates. In some implementations, when using Type 3 training collaboration, a model update process can be used. Type 1 training collaboration can use model transfer.

[0054] Figure 6 Example signaling flow diagram 602 illustrates a B1-type model update according to some implementation schemes. The illustrated implementation scheme is a B1-type model update initiated by the network (e.g., network node 604). For NW-first training, when network node 604 can train the updated model 608, the new training dataset 610 is transferred to UE 606.

[0055] UE 606 can use the new training dataset 610 to train encoder 612 offline. UE 606 can indicate to network node 604 that 614 supports updated model IDs. Furthermore, UE 606 can indicate to network node 604 the expectation of updating the model IDs / dataset IDs supported by UE 606. This indication can be made through a UE capability report, or the UAI can indicate the new model ID / dataset ID via RRC signaling.

[0056] Network node 604 can transmit activation / deactivation message 616 to UE 606 to activate or deactivate the model. UE 606 and network node 604 can continue the remaining lifecycle management process (e.g., activation, monitoring, deactivation, or switching of the model).

[0057] Figure 7 Example signal flow diagram 702 illustrates a B2-type model update according to some implementation schemes. The illustrated implementation scheme is a B2-type model update initiated by UE 706. As an example, for UE-priority training, UE 706 can train the updated model 708 and transmit the new training dataset 710 to network node 704.

[0058] Network node 704 can use the new training dataset 710 to train encoder 712 offline. Network node 704 can indicate to UE 706 that 714 supports updated model IDs. Network node 704 can also indicate to UE 706 the expected updated model IDs / dataset IDs supported by network node 704. This indication can be made via RRC configuration / downlink control information (DCI) and / or MAC CE.

[0059] Network node 704 can transmit activation / deactivation message 716 to UE 706 to activate or deactivate the model. UE 706 and network node 704 can continue the remaining lifecycle management process (e.g., activation, monitoring, deactivation, or switching of the model).

[0060] In some implementations, sub-models can be used in conjunction with pattern transfer and model update processes. For direct AI / ML localization, evaluation results indicate that fine-tuning / retraining the previous model using a dataset from the new deployment scenario improves its performance in the new deployment scenario. After fine-tuning / retraining the previous model using a dataset from the new deployment scenario, the performance of the updated model may degrade for the previous deployment scenario in which the previous model was originally trained (e.g., previous clutter parameter settings).

[0061] Therefore, in some implementations, the wireless communication system can retain the original model in case a switchback to the original model is needed. For this purpose, a sub-model / child-model framework can be created. The framework can use the original model ID as the parent model ID and a fine-tuned / retrained model as the child model associated with the parent model. The UE can indicate the number of sub-models it can support as its capability. In relation to Figures 1 to 6 In the described implementation, the UE / network can indicate a new sub-model ID during transmission. In some implementations, the parent model can be retained by default. In some implementations, associated signaling can be used to indicate whether the parent model should be retained.

[0062] Figure 8 A sub-model 802 for incorporating model transfer during fine-tuning is illustrated according to some implementation schemes. In model fine-tuning, layers close to the model input can be frozen, and layers close to the model output can be updated. For example, model 806 may have six layers, and the model ID is set to X. Sub-model 802 can be created by freezing the first four layers 808 closer to the input and modifying the two layers 804 closer to the model output.

[0063] In some implementations, model transfers or model updates can indicate which layers to change and how to change them. A sub-model ID indicating a change to a specific layer can be defined. For example, a sub-ID set to Y could indicate sub-model 802 whose two layers 808 closest to the output have been changed. Model identification and transfers can involve identifying / transferring only the identified / configured layers. For example, a model transfer or update could provide a mapping indicating layers that have been changed.

[0064] Currently, some conditions applied on the UE side are not provided to the network side, and vice versa. Some of these conditions can be used to train the model more accurately. Therefore, Figure 9 and Figure 10 This example demonstrates how to include additional conditions in the model identifier.

[0065] Specifically, Figure 9 A signal flow diagram 902 illustrates a model identifier for network-side additional conditions according to some implementation schemes. Network-side additional conditions can refer to network-side specific implementations transparent to the UE. For example, for the CSI compression use case, network antenna-to-port virtualization can be a network-side additional condition. Furthermore, for beam management, the network-side beam pattern used for transmitting synchronization signal blocks (SSBs) and channel state information reference signals (CSI-RS) can be a network-side additional condition.

[0066] Signal flow diagram 902 illustrates a B1 type initiated by network node 904. Network node 904 can determine a model ID 906 corresponding to its own additional conditions. The model ID can be a logical ID used to classify the dataset. Network node 904 can create new model IDs to use as identifiers for different condition configurations. For example, if network node 904 changes the side beam pattern, the network node can associate the new side beam pattern with a new model ID. Network node 904 can use the model ID to trigger the data collection process 908.

[0067] UE 904 can collect data 912 and train the model offline. UE 904 can collect data with the same ID and train the model offline. In some implementations, the UE can train the model to have generalization performance, thus enabling it to work with multiple model IDs and / or datasets.

[0068] In some implementations, network node 904 can transmit configuration 914 of a use case with a model ID to UE 910 for inference. UE 910 can indicate 916 to network node 904 whether the model ID is supported. In other implementations, UE 910 can report supported model IDs to network node 904 in UE capabilities or UAI, and network node 904 can configure / activate the model ID for inference.

[0069] Network node 904 can transmit activation / deactivation messages 918 to UE 910 to activate or deactivate the model. UE 910 and network node 904 can continue the remaining lifecycle management process (e.g., activation, monitoring, deactivation, or switching of the model).

[0070] Figure 10 A signal flow diagram 1002 illustrating model identification for UE-side additional conditions according to some implementation schemes is provided. UE-side additional conditions may refer to specific UE-side implementations that are transparent to the network. For example, the placement of the UE antenna on a telephone can be a UE-side additional condition. Furthermore, the UE-side beam pattern used for beam management can be a UE-side additional condition.

[0071] Signal flow diagram 1002 illustrates a B2 type initiated by UE 1004. For network-side data collection, network node 1006 can trigger data collection process 1008. UE 1004 can determine 1010 UE-side additional conditions and the corresponding model ID. In the data report 1012 from UE 1004 to network node 1006, UE 1004 can transmit the model ID corresponding to the UE-side additional conditions. Network node 1006 can collect data with the same ID and train 1014 the model offline. Network node 1006 can train the model to have generalization performance, thus enabling it to work with multiple model IDs and / or datasets.

[0072] In some implementations, network node 1006 may be configured with model ID 1016 to allow the UE to determine which model matches the additional conditions on the UE side. The UE may indicate the model ID corresponding to 1018, and the network may perform inference.

[0073] In some implementations, UE 1004 may report the supported model IDs to network node 1006 in a UE Capability Report or UAI. Network node 1006 can then use the model corresponding to the model ID for inference.

[0074] Network node 1006 can transmit activation / deactivation message 1020 to UE 1004 to activate or deactivate the model. UE 1004 and network node 1006 can continue the remaining lifecycle management process (e.g., activation, monitoring, deactivation, or switching of the model).

[0075] Figure 11 A method 1100 for a UE according to an embodiment of this document is illustrated. The illustrated method 1100 includes training 1102 a new model for wireless communication with a network node. Method 1100 also includes transmitting 1104 an indication to the network node including a new model ID associated with the new model. Method 1100 further includes receiving 1106 an uplink grant from the network node including a request for model transmission. Method 1100 further includes transmitting 1108 the new model along with the new model ID to the network node. Method 1100 also includes receiving 1110 a message from the network node for activating the new model.

[0076] In some implementations of method 1100, the new model includes a model structure known to the network nodes. In some such implementations, the new model ID includes a version number, and the new model is signaled via the version ID. In other such implementations, the new model ID is associated with a previous model ID in signaling, and the previous model ID implicitly indicates a known model structure. Some such implementations also include adding a value label to the previous model ID to generate the new model ID.

[0077] In some implementations, method 1100 also includes transmitting metadata to network nodes to provide more information about the new model.

[0078] In some implementations of method 1100, the model is an unknown model.

[0079] In some implementations, method 1100 further includes: training a new model to generate an updated model; transmitting a new training dataset for the updated model to a network node; and receiving an instruction from the network node to update the new model ID or dataset ID. In some such implementations, when a new model is updated, the UE retains both the original version of the new model and the sub-models that include the updated model.

[0080] In some implementations, method 1100 also includes associating additional conditions on the UE side with a new model ID.

[0081] In some implementations, method 1100 also includes associating network-side additional conditions with a new model ID, wherein the new model ID associated with the network-side additional conditions is part of the data collection process.

[0082] The embodiments contemplated herein include an apparatus comprising components for performing one or more elements of method 1100. This apparatus may be, for example, a UE (such as the wireless device 1402 described herein as a UE).

[0083] The embodiments contemplated herein include one or more non-transitory computer-readable media comprising instructions that, when executed by one or more processors of an electronic device, cause the electronic device to perform one or more elements of method 1100. The non-transitory computer-readable media may be, for example, the memory of a UE (such as memory 1406 of a wireless device 1402 as a UE as described herein).

[0084] The embodiments contemplated herein include an apparatus comprising logic components, modules, or circuitry for performing one or more elements of method 1100. This apparatus may be, for example, a UE (such as the wireless device 1402 described herein as a UE).

[0085] The embodiments contemplated herein include an apparatus comprising: one or more processors and one or more computer-readable media including instructions that, when executed by the one or more processors, cause the one or more processors to perform one or more elements of method 1100. The apparatus may be, for example, a UE (such as the wireless device 1402 described herein as a UE).

[0086] The implementation scheme envisioned herein includes a signal as described in or related to one or more elements of method 1100.

[0087] The embodiments contemplated herein include a computer program or computer program product comprising instructions, wherein execution of the program by a processor will cause the processor to perform one or more elements of method 1100. The processor may be a processor of a UE (such as processor 1404 of wireless device 1402 as a UE as described herein). These instructions may be, for example, located in the processor and / or on the memory of the UE (such as memory 1406 of wireless device 1402 as a UE as described herein).

[0088] Figure 12 A method 1200 for a network node according to an embodiment of this document is illustrated. The illustrated method 1200 includes training 1202 a new model for wireless communication with a UE. Method 1200 also includes transmitting 1204 an announcement to the UE including a new model ID associated with the new model. Method 1200 also includes receiving 1206 a request from the UE to download the new model. Method 1200 also includes transmitting 1208 the new model along with the new model ID to the UE. Method 1200 also includes receiving 1210 an indication from the UE that the UE is ready to use the new model for inference. Method 1200 also includes transmitting 1212 a message to the UE to activate the new model.

[0089] In some implementations of method 1200, the new model includes a model structure known to the network nodes. In some such implementations, the new model ID includes a version number, and the new model is signaled via the version ID. In other such implementations, the new model ID is associated with a previous model ID in signaling, and the previous model ID implicitly indicates a known model structure. Some such implementations also include adding a value label to the previous model ID to generate the new model ID.

[0090] In some implementations, method 1200 also includes transmitting metadata to the UE to provide more information about the new model.

[0091] In some implementations of method 1200, the model is an unknown model.

[0092] In some implementations, method 1200 further includes: training a new model to generate an updated model; transmitting a new training dataset for the updated model to the UE; and receiving an indication from the UE that supports the updated model. In some such implementations, when the new model is updated, both the original version of the new model and a sub-model including the updated model are stored.

[0093] In some implementations, method 1200 also includes associating network-side additional conditions with a new model ID.

[0094] In some implementations, method 1200 also includes associating additional conditions on the UE side with a new model ID.

[0095] The embodiments contemplated herein include an apparatus comprising components for performing one or more elements of method 1200. This apparatus may be, for example, a base station (such as network device 1418 as a base station described herein).

[0096] The embodiments contemplated herein include one or more non-transitory computer-readable media comprising instructions that, when executed by one or more processors of an electronic device, cause the electronic device to perform one or more elements of method 1200. The non-transitory computer-readable medium may be, for example, the memory of a base station (such as memory 1422 of network device 1418 as a base station as described herein).

[0097] The embodiments contemplated herein include an apparatus comprising logic components, modules, or circuitry for performing one or more elements of method 1200. This apparatus may be, for example, an apparatus for a base station (such as network device 1418 as a base station as described herein).

[0098] The embodiments contemplated herein include an apparatus comprising: one or more processors and one or more computer-readable media including instructions that, when executed by the one or more processors, cause the one or more processors to perform one or more elements of method 1200. The apparatus may be, for example, an apparatus for a base station (such as network device 1418 as a base station described herein).

[0099] The implementation scheme envisioned herein includes a signal as described in or related to one or more elements of method 1200.

[0100] The embodiments contemplated herein include a computer program or computer program product comprising instructions, wherein execution by a processing element causes the processing element to perform one or more elements of method 1200. The processor may be a processor of a base station (such as processor 1420 of network device 1418 as a base station as described herein). These instructions may, for example, be located in the processor and / or in the memory of the base station (such as memory 1422 of network device 1418 as a base station as described herein).

[0101] Figure 13 An example architecture of a wireless communication system 1300 according to an embodiment disclosed herein is illustrated. The following description is provided for the example wireless communication system 1300, which operates in conjunction with the LTE system standard and / or 5G or NR system standard provided by the 3GPP technical specifications.

[0102] like Figure 13 As shown, the wireless communication system 1300 includes UE 1302 and UE 1304 (but any number of UEs may be used). In this example, UE 1302 and UE 1304 are exemplified as smartphones (e.g., handheld touchscreen mobile computing devices capable of connecting to one or more cellular networks), but may also include any mobile or non-mobile computing device configured for wireless communication.

[0103] UE 1302 and UE 1304 can be configured to be communicatively coupled to RAN 1306. In an implementation, RAN 1306 can be NG-RAN, E-UTRAN, etc. UE 1302 and UE 1304 utilize connections (or channels) with RAN 1306 (shown as connection 1308 and connection 1310, respectively), where each connection includes a physical communication interface. RAN 1306 may include one or more base stations (such as base station 1312 and base station 1314) implementing connection 1308 and connection 1310.

[0104] In this example, Connection 1308 and Connection 1310 are air interfaces that implement this type of communication coupling and can conform to the RAT used by RAN 1306, such as LTE and / or NR, for example.

[0105] In some implementations, UE 1302 and UE 1304 may also exchange communication data directly via sidelink interface 1316. UE 1304 is shown configured to access an access point (shown as AP 1318) via connection 1320. By way of example, connection 1320 may include a local wireless connection, such as a connection conforming to any IEEE 802.11 protocol, while AP 1318 may include Wi-Fi. ® Router. In this example, AP 1318 can connect to another network (e.g., the Internet) without using CN 1324.

[0106] In the implementation, UE 1302 and UE 1304 may be configured to communicate with each other or with base station 1312 and / or base station 1314 on a multi-carrier communication channel using orthogonal frequency division multiplexing (OFDM) communication signals according to various communication technologies, such as but not limited to orthogonal frequency division multiple access (OFDMA) communication technology (e.g., for downlink communication) or single-carrier frequency division multiple access (SC-FDMA) communication technology (e.g., for uplink and ProSe or sidelink communication), but the scope of the implementation is not limited in this respect. The OFDM signal may include multiple orthogonal subcarriers.

[0107] In some implementations, all or some of the base stations in base station 1312 or base station 1314 may be implemented as one or more software entities running on a server computer as part of a virtual network. Furthermore, or in other implementations, base station 1312 or base station 1314 may be configured to communicate with each other via interface 1322. In implementations where the wireless communication system 1300 is an LTE system (e.g., when CN 1324 is an EPC), interface 1322 may be an X2 interface. This X2 interface may be defined between two or more base stations (e.g., two or more eNBs, etc.) connected to the EPC and / or between two eNBs connected to the EPC. In implementations where the wireless communication system 1300 is an NR system (e.g., when CN 1324 is a 5GC), interface 1322 may be an Xn interface. The Xn interface is defined between two or more base stations (e.g., two or more gNBs, etc.) connected to the 5GC, between a base station 1312 (e.g., a gNB) connected to the 5GC and an eNB, and / or between two eNBs connected to the 5GC (e.g., CN 1324).

[0108] RAN 1306 is shown communicatively coupled to CN 1324. CN 1324 may include one or more network elements 1326 configured to provide various data and telecommunications services to customers / subscribers (e.g., users of UE 1302 and UE 1304) connected to CN 1324 via RAN 1306. Components of CN 1324 may be implemented in a single physical device or a separate physical device including components for reading and executing instructions from machine-readable or computer-readable media (e.g., non-transitory machine-readable storage media).

[0109] In the implementation scheme, CN 1324 may be an EPC, and RAN 1306 may be connected to CN 1324 via S1 interface 1328. In the implementation scheme, S1 interface 1328 may be divided into two parts: an S1 user plane (S1-U) interface, which carries service data between base station 1312 or base station 1314 and the serving gateway (S-GW); and an S1-MME interface, which is the signaling interface between base station 1312 or base station 1314 and the mobility management entity (MME).

[0110] In the implementation scheme, CN 1324 may be a 5GC, and RAN 1306 may be connected to CN 1324 via NG interface 1328. In the implementation scheme, NG interface 1328 may be divided into two parts: an NG user plane (NG-U) interface, which carries service data between base station 1312 or base station 1314 and user plane function (UPF); and an S1 control plane (NG-C) interface, which is the signaling interface between base station 1312 or base station 1314 and access and mobility management function (AMF).

[0111] Generally, application server 1330 may be an element that provides Internet Protocol (IP) bearer resources (e.g., packet-switched data services) for use with CN 1324. Application server 1330 may also be configured to support one or more communication services (e.g., VoIP sessions, group communication sessions, etc.) for UE 1302 and UE 1304 via CN 1324. Application server 1330 can communicate with CN 1324 via IP communication interface 1332.

[0112] Figure 14A system 1400 for performing signaling 1434 between a wireless device 1402 and a network device 1418, according to an embodiment disclosed herein, is illustrated. System 1400 may be part of a wireless communication system as described herein. Wireless device 1402 may be, for example, a UE of a wireless communication system. Network device 1418 may be, for example, a base station (e.g., an eNB or gNB) of a wireless communication system.

[0113] Wireless device 1402 may include one or more processors 1404. Processor 1404 is executable instructions that enable various operations of wireless device 1402 to be performed as described herein. Processor 1404 may include one or more baseband processors, which are implemented using, for example, a central processing unit (CPU), digital signal processor (DSP), application-specific integrated circuit (ASIC), controller, field-programmable gate array (FPGA) device, another hardware device, firmware device, or any combination thereof configured to perform the operations described herein.

[0114] Wireless device 1402 may include memory 1406. Memory 1406 may be a non-transitory computer-readable storage medium that stores instructions 1408, which may include, for example, instructions executed by processor 1404. Instructions 1408 may also be referred to as program code or a computer program. Memory 1406 may also store data used by processor 1404 and results calculated by the processor.

[0115] Wireless device 1402 may include one or more transceivers 1410, which may include radio frequency (RF) transmitter circuitry and / or receiver circuitry that uses antenna 1412 of wireless device 1402 to facilitate signaling (e.g., signaling 1434) to and / or from wireless device 1402 and other devices (e.g., network device 1418) in accordance with a corresponding RAT.

[0116] Wireless device 1402 may include one or more antennas 1412 (e.g., one, two, four or more). In embodiments with multiple antennas 1412, wireless device 1402 may utilize spatial diversity of such multiple antennas 1412 to transmit and / or receive multiple different data streams on the same time-frequency resource. This behavior may be referred to as, for example, multiple-input multiple-output (MIMO) behavior (referring to multiple antennas used at each of the transmitting and receiving devices to implement this aspect). MIMO transmission by wireless device 1402 may be achieved according to pre-decoding (or digital beamforming) applied at wireless device 1402, which multiplexes data streams across antennas 1412 based on known or assumed channel characteristics, such that each data stream is received with appropriate signal strength relative to the others at a desired location in the spatial domain (e.g., the location of the receiver associated with that data stream). Some implementations may use a single-user MIMO (SU-MIMO) approach (where all data streams are directed to a single receiver) and / or a multi-user MIMO (MU-MIMO) approach (where individual data streams may be directed to individual (different) receivers at different locations in the airspace).

[0117] In some implementations with multiple antennas, wireless device 1402 may implement analog beamforming technology, whereby the phase of the signal transmitted by antenna 1412 is relatively adjusted so that the (joint) transmission of antenna 1412 can be directed (this is sometimes referred to as beam control).

[0118] Wireless device 1402 may include one or more interfaces 1414. Interface 1414 can be used to provide input to or output to wireless device 1402. For example, wireless device 1402 as a UE may include interface 1414, such as a microphone, speaker, touchscreen, and buttons, to allow a user of the UE to make inputs and / or outputs to the UE. Other interfaces of such a UE may consist of transmitters, receivers, and other circuitry that allow the UE to communicate with other devices (e.g., in addition to the transceiver 1410 / antenna 1412 already described), and may be based on known protocols (e.g., Wi-Fi). ® and Bluetooth ® (etc.) to perform the operation.

[0119] Wireless device 1402 may include a model transmission and update module 1416. The model transmission and update module 1416 may be implemented via hardware, software, or a combination thereof. For example, the model transmission and update module 1416 may be implemented as a processor, circuitry, and / or instructions 1408 stored in memory 1406 and executed by processor 1404. In some examples, the model transmission and update module 1416 may be integrated within processor 1404 and / or transceiver 1410. For example, the model transmission and update module 1416 may be implemented by a combination of software components and hardware components (e.g., logic gates and circuitry) within processor 1404 or transceiver 1410 (e.g., executed by a DSP or general-purpose processor).

[0120] The model transfer and update module 1416 can be used in various aspects of this disclosure, for example, Figures 1 to 13 All aspects. The model transfer and update module 1416 is configured to perform the process of transferring and updating AI / ML models.

[0121] Network device 1418 may include one or more processors 1420. Processor 1420 is executable instructions to perform various operations of network device 1418 as described herein. Processor 1420 may include one or more baseband processors, which are implemented using, for example, a CPU, DSP, ASIC, controller, FPGA device, another hardware device, firmware device, or any combination thereof configured to perform the operations described herein.

[0122] Network device 1418 may include memory 1422. Memory 1422 may be a non-transitory computer-readable storage medium that stores instructions 1424, which may include, for example, instructions executed by processor 1420. Instructions 1424 may also be referred to as program code or a computer program. Memory 1422 may also store data used by processor 1420 and results calculated by the processor.

[0123] Network device 1418 may include one or more transceivers 1426, which may include RF transmitter circuitry and / or receiver circuitry that uses the antenna 1428 of network device 1418 to facilitate signaling (e.g., signaling 1434) to and / or from network device 1418 and other devices (e.g., wireless device 1402) according to the corresponding RAT.

[0124] Network device 1418 may include one or more antennas 1428 (e.g., one, two, four or more). In embodiments having multiple antennas 1428, network device 1418 may perform MIMO, digital beamforming, analog beamforming, beam control, etc., as described.

[0125] Network device 1418 may include one or more interfaces 1430. Interface 1430 may be used to provide input to or output to network device 1418. For example, network device 1418 as a base station may include interface 1430 consisting of transmitters, receivers and other circuitry (e.g., in addition to the transceiver 1426 / antenna 1428 already described), which enable the base station to communicate with other equipment in the core network and / or enable the base station to communicate with external networks, computers and databases, etc., for the purpose of performing operations, management and maintenance of the base station or other equipment operatively connected to the base station.

[0126] Network device 1418 may include a model transfer and update module 1432. The model transfer and update module 1432 may be implemented via hardware, software, or a combination thereof. For example, the model transfer and update module 1432 may be implemented as a processor, circuitry, and / or instructions 1424 stored in memory 1422 and executed by processor 1420. In some examples, the model transfer and update module 1432 may be integrated within processor 1420 and / or transceiver 1426. For example, the model transfer and update module 1432 may be implemented by a combination of software components and hardware components (e.g., logic gates and circuitry) within processor 1420 or transceiver 1426 (e.g., executed by a DSP or general-purpose processor).

[0127] The model transfer and update module 1432 can be used in various aspects of this disclosure, for example, Figures 1 to 13 All aspects. The model transfer and update module 1432 is configured to perform the process of transferring and updating AI / ML models.

[0128] For one or more embodiments, at least one of the components illustrated in one or more of the foregoing figures may be configured to perform one or more operations, techniques, processes, and / or methods as described herein. For example, a baseband processor as described herein in conjunction with one or more of the foregoing figures may be configured to operate according to one or more of the examples illustrated herein. Similarly, circuitry associated with a UE, base station, network element, etc., as described above in conjunction with one or more of the foregoing figures may be configured to operate according to one or more of the examples illustrated herein.

[0129] Unless otherwise expressly stated, any of the embodiments described above may be combined with any other embodiment (or combination of embodiments). The foregoing description of one or more specific embodiments provides illustrative and descriptive information, but is not intended to be exhaustive or to limit the scope of the embodiments to the precise form disclosed. In light of the teachings above, modifications and variations are possible, or modifications and variations may be derived from practice with various embodiments.

[0130] Implementations and specific embodiments of the systems and methods described herein may include various operations embodied in machine-executable instructions to be executed by a computer system. The computer system may include one or more general-purpose or special-purpose computers (or other electronic devices). The computer system may include hardware components, including specific logical parts for performing the operations; or may include a combination of hardware, software, and / or firmware.

[0131] It should be recognized that the systems described herein include descriptions of specific implementations. These implementations may be combined into a single system, partially integrated into other systems, divided into multiple systems, or otherwise partitioned or combined. Furthermore, it is conceivable to use parameters, attributes, aspects, etc., of one implementation in one implementation. For clarity, these parameters, attributes, aspects, etc., are described only in one or more implementations, and it should be recognized that, unless expressly stated herein, these parameters, attributes, aspects, etc., may be combined with or substituted for parameters, attributes, aspects, etc., of another implementation.

[0132] As is widely recognized, the use of personally identifiable information should comply with privacy policies and practices that are generally accepted to meet or exceed industry or governmental requirements for protecting user privacy. Specifically, personally identifiable information data should be managed and processed to minimize the risk of unintentional or unauthorized access or use, and the nature of authorized use should be clearly explained to users.

[0133] Although the foregoing has been described in considerable detail for clarity, it will be apparent that certain changes and modifications can be made without departing from the principles of the invention. It should be noted that there are many alternative ways to implement both the processes and apparatus described herein. Therefore, embodiments of the invention should be considered illustrative rather than restrictive, and this specification is not limited to the details given herein, but can be modified within the scope and equivalents of the appended claims.

Claims

1. A method for user equipment (UE), the method comprising: Train a new model for wireless communication with network nodes; Transmit to the network node an indication including a new model identifier (ID) associated with the new model; Receive uplink permission from the network node, including a request for model transmission; The new model, along with its ID, is transmitted to the network node. as well as Receive a message from the network node to activate the new model.

2. The method according to claim 1, wherein the new model includes a model structure known to the network nodes.

3. The method of claim 2, wherein the new model ID includes a version number, and wherein the new model is signaled via the version ID.

4. The method of claim 2, wherein the new model ID is associated with a previous model ID in the signaling, and wherein the previous model ID implicitly indicates a known model structure.

5. The method of claim 2, further comprising adding a value label to the previous model ID to generate the new model ID.

6. The method of claim 1, further comprising transmitting meta-information to the network node to provide more information about the new model.

7. The method according to claim 1, wherein the new model is an unknown model.

8. The method according to claim 1, further comprising: The new model is trained to generate an updated model; Transmit the new training dataset for the updated model to the network node; as well as Receive a second instruction from the network node for updating the new model ID or dataset ID.

9. The method of claim 8, wherein when the new model is updated, the UE retains both the original version of the new model and the sub-model that includes the updated model.

10. The method of claim 1, further comprising associating UE-side additional conditions with the new model ID.

11. The method of claim 1, further comprising associating network-side additional conditions with the new model ID, wherein the new model ID associated with the network-side additional conditions is part of a data collection process.

12. A method for a network node, the method comprising: Train a new model for wireless communication with user equipment (UE); The UE is sent an announcement including a new model identifier (ID) associated with the new model; Receive a request from the UE to download the new model; The new model, along with its ID, is transmitted to the UE. Receive from the UE an indication that the UE is ready to use the new model for inference; as well as A message for activating the new model is transmitted to the UE.

13. The method of claim 12, wherein the new model comprises a model structure known to the network nodes.

14. The method of claim 13, wherein the new model ID includes a version number, and wherein the new model is signaled via the version ID.

15. The method of claim 13, wherein the new model ID is associated with a previous model ID in signaling, and wherein the previous model ID implicitly indicates a known model structure.

16. The method of claim 13, further comprising adding a value label to a previous model ID to generate the new model ID.

17. The method of claim 12, further comprising transmitting metadata to the UE to provide more information about the new model.

18. The method of claim 12, wherein the new model is an unknown model.

19. The method according to claim 12, further comprising: The new model is trained to generate an updated model; Transmit a new training dataset for the updated model to the UE; as well as The UE receives an indication of the model that supports the update.

20. The method of claim 19, wherein when the new model is updated, both the original version of the new model and a sub-model including the updated model are stored.

21. The method of claim 12, further comprising associating network-side additional conditions with the new model ID.

22. The method of claim 12, further comprising associating UE-side additional conditions with the new model ID.

23. An apparatus comprising components for performing the method according to any one of claims 1 to 21.

24. A computer-readable medium comprising instructions that, when executed by one or more processors of an electronic device, cause the electronic device to perform the method according to any one of claims 1 to 21.

25. An apparatus comprising a logic component, module, or circuitry for performing the method according to any one of claims 1 to 21.

Images