Triggering artificial intelligence / machine learning training in network data analysis function

Abstract

And triggering artificial intelligence / machine learning training in a network data analysis function is provided. A method for triggering artificial intelligence / machine learning training in a network data analysis function may include obtaining at least one machine learning model for training or retraining based on measurement data of a network. The method may also include determining that data collection is required prior to training or retraining the at least one machine learning model, and receiving one or more measurement reports including at least one data set from the data collection. The method may also include determining whether additional assistance information from the one or more network devices requires training or retraining. At least one machine learning model may be trained or retrained based on all collected data sets, which include at least one data set from the data collection.

Description

[0001] Related Applications

[0002] This application claims priority to U.S. Provisional Application No. 63 / 531,966, filed on August 10, 2023, which is incorporated herein by reference in its entirety. Technical Field

[0003] Some example embodiments may generally relate to mobile or wireless telecommunication systems, such as Long Term Evolution (LTE) or fifth generation (5G) New Radio (NR) access technology, or sixth generation (6G technology), or other communication systems. For example, certain example embodiments may relate to triggering artificial intelligence / machine learning training in a network data analysis function. Background Art

[0004] Examples of mobile or wireless telecommunication systems may include Universal Mobile Telecommunications System (UMTS) Terrestrial Radio Access Network (UTRAN), Long Term Evolution (LTE), Evolved UTRAN (E-UTRAN), LTE Advanced (LTE-A), MulteFire, LTE-A Pro and / or fifth generation (5G) radio access technology or new radio (NR) access technology. The fifth generation (5G) wireless system refers to the next generation (NG) radio system and network architecture. 5G network technology is mainly based on new radio (NR) technology, but 5G (or NG) networks can also be built on E-UTRAN radio. It is estimated that NR can provide bit rates of about 10-20Gbit / s or higher, and can support at least enhanced mobile broadband (eMBB) and ultra-reliable low latency communication (URLLC) and massive machine type communication (mMTC). NR is expected to provide extremely wide bandwidth and ultra-robust, low latency connectivity and large-scale networking to support the Internet of Things (IoT). Summary of the invention

[0005] Various exemplary embodiments may provide an apparatus 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 obtain at least one machine learning model for training or retraining based on network measurement data from at least one network entity. The apparatus may also be caused to determine that data collection is required before training or retraining at least one machine learning model, and receive one or more measurement reports including at least one data set from data collection from a network function. The apparatus may also be caused to determine whether additional auxiliary information from one or more network devices requires training or retraining, and train or retrain at least one machine learning model based on all collected data sets, including at least one data set from data collection.

[0006] Certain exemplary embodiments may provide an apparatus, the apparatus comprising at least one processor and at least one memory, the at least one memory storing instructions that, when executed by the at least one processor, cause the apparatus to at least determine that at least one machine learning model needs to be updated, and send a request to a network entity to train or retrain at least one machine learning model. The apparatus may also provide at least one measurement report indicating the performance of a user equipment to the network entity, and receive from the network entity a trained or retrained machine learning model to be executed on the apparatus.

[0007] Certain exemplary embodiments may provide a method comprising obtaining, by a device, from a network entity at least one machine learning model for training or retraining based on network measurement data. The method may also include determining that data collection is required before training or retraining at least one machine learning model, and receiving one or more measurement reports from a network function including at least one data set from data collection. The method may also include determining whether additional auxiliary information from one or more network devices requires training or retraining, and training or retraining at least one machine learning model based on all collected data sets, including at least one data set from data collection.

[0008] Certain exemplary embodiments may provide a method, the method comprising determining, by an apparatus, that at least one machine learning model needs to be updated, and sending a request to a network entity to train or retrain the at least one machine learning model. The method may also include providing at least one measurement report indicating user equipment performance to the network entity, and receiving from the network entity a trained or retrained machine learning model to be executed on the apparatus.

[0009] Various exemplary embodiments may provide an apparatus comprising a component for obtaining from a network entity at least one machine learning model for training or retraining based on network measurement data. The apparatus may also include a component for determining that data collection is required before training or retraining at least one machine learning model, and a component for receiving from a network function one or more measurement reports including at least one data set from data collection. The apparatus may also include a component for determining whether additional auxiliary information from one or more network devices requires training or retraining, and a component for training or retraining at least one machine learning model based on all collected data sets, including at least one data set from data collection.

[0010] Some exemplary embodiments may provide an apparatus including means for determining, by the apparatus, that at least one machine learning model needs to be updated, and means for sending a request to a network entity to train or retrain the at least one machine learning model. The apparatus may also include means for providing at least one measurement report indicating user equipment performance to the network entity, and means for receiving from the network entity a trained or retrained machine learning model to be executed on the apparatus.

[0011] Certain exemplary embodiments may provide a non-transitory computer-readable medium comprising program instructions that, when executed by an apparatus, cause the apparatus to obtain at least one machine learning model for training or retraining based on network measurement data from at least one network entity. The apparatus may also be caused to determine that data collection is required before training or retraining at least one machine learning model, and receive one or more measurement reports including at least one data set from data collection from a network function. The apparatus may also be caused to determine whether additional auxiliary information from one or more network devices requires training or retraining, and train or retrain at least one machine learning model based on all collected data sets, including at least one data set from data collection.

[0012] Various exemplary embodiments may provide a non-transitory computer-readable medium including program instructions that, when executed by an apparatus, cause the apparatus to at least determine that at least one machine learning model needs to be updated and send a request to a network entity to train or retrain at least one machine learning model. The apparatus may also provide at least one measurement report indicating the performance of a user equipment to the network entity and receive a trained or retrained machine learning model to be executed on the apparatus from the network entity.

[0013] Some exemplary embodiments may provide one or more computer programs including instructions stored thereon for performing one or more methods described herein. Some exemplary embodiments may also provide one or more devices including one or more circuits configured to perform one or more methods described herein. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] In order to properly understand the example embodiments, reference should be made to the accompanying drawings, as follows:

[0015] Figure 1 An example of a signaling diagram of a signaling-based MDT procedure is shown;

[0016] Figure 2 An example of a signaling diagram for configuring signaling-based MDT is shown;

[0017] Figure 3 shows an example of a flow chart for triggering AI / ML training or retraining using MDT according to various example embodiments;

[0018] Figure 4A shows a signaling diagram of one or more processes for triggering training or retraining of a model according to some example embodiments;

[0019] Figure 4B A signaling diagram illustrating one or more processes for triggering training or retraining of a model according to some example embodiments is Figure 4A continuation of

[0020] Figure 5 illustrates an example of a signaling diagram for one or more procedures for direct data forwarding and configuration reporting according to certain example embodiments;

[0021] Figure 6 An example of a flow chart illustrating a method according to various example embodiments;

[0022] Figure 7 shows an example of a flowchart of another method according to some exemplary embodiments; and

[0023] Figure 8 A set of apparatuses according to various exemplary embodiments is shown. DETAILED DESCRIPTION

[0024] It will be readily appreciated that the components of certain example embodiments as generally described and illustrated in the figures herein may be arranged and designed in a variety of different configurations. Below is a detailed description of some example embodiments of systems, methods, apparatus, and non-transient computer program products for triggering an artificial intelligence / machine learning (AI / ML) training process in a network data analysis function (NWDAF). Although the devices discussed below and shown in the figures refer to 5G or next-generation NodeB (gNB) devices, network devices, and user equipment (UE), the present disclosure is not limited to the gNBs, UEs, and network elements mentioned herein. For example, the following description may also apply to any type of network device or element and UE.

[0025] 5G / NR networks may have the capability to support various communication services such as the Internet of Things (IoT) and enhanced mobile broadband (eMBB). Large amounts of data related to network and service events and status may need to be analyzed and processed.

[0026] Data collection may be particularly useful for implementing artificial intelligence or machine learning (AI / ML) in the NR air interface. Certain existing or legacy data collection processes in 5G networks may not be sufficient to train AI / ML models in the gNB or radio access network (RAN) and perform subsequent model updates / fine-tuning. MDT can be used to collect RAN-related measurements from one or more UEs and can report the measurements to the gNB via, for example, measurement reports. MDT may include signaling-based MDT and / or management-based MDT, both of which may require the user to consent to the MDT before activating the MDT functionality due to privacy and legal obligations. The owner or operator of the device(s) performing the MDT process may need to collect this user consent before performing the MDT process. Information related to user consent may be accessible as part of the subscription data and may be stored in a unified data management (UDM) database.

[0027] Figure 1 An example of a signaling diagram of a signaling-based MDT procedure that facilitates accessing or obtaining user consent is shown. Figure 1 The network configuration in the example may include a management service (MaS) function 101, a UDM 102, an access and mobility function (AMF) 103, and a base station or network node, such as a gNB 104. The signaling-based MDT process may include: at 110, the MaS 101 may provide a user consent provision to the UDM 102, and at 120, the UDM 102 may store the user consent for a specific user.

[0028] At 130, MaS 101 may provide International Mobile Subscriber Identity (IMSI) and / or Subscription Permanent Identifier (SUPI) based Minimization of Drive Test (MDT) activation to UDM 102; and at 140, UDM 102 may determine whether user consent is available. At 150, if it is determined that user consent is not available or has not been given, UDM 102 may reject the MDT activation and notify MaS 101 of the rejection. At 155, if it is determined that user consent is available, UDM 102 may perform MDT activation with AMF 103 at 160. At 170, gNB 104 may also perform MDT activation, and AMF 103 may send an activation message.

[0029] Figure 2 An example of a signaling diagram for configuring signaling-based MDT in a UE after the user confirms consent is shown. Figure 2The network configuration in the example may include MaS 201, UDM 202, AMF 203, a base station or network node (such as gNB 204), and UE 205. The process for configuring signaling-based MDT in UE 205 may include that at 210, UDM 202 may perform MDT activation with AMF 203, and at 220, AMF 203 may perform MDT activation with gNB 204.

[0030] At 230, gNB 204 may perform a radio resource control (RRC) reconfiguration with UE 205, which may include reporting the configuration of the MDT. At 240, UE 205 may complete the RRC reconfiguration with gNB 204.

[0031] Various exemplary embodiments may provide advantages and / or improvements to conventional MDT configuration procedures by, for example, implementing NWDAF and triggering data collection of RAN measurements through MDT for offline AI / ML model training or retraining.

[0032] Certain exemplary embodiments may provide for triggering offline training or retraining for UE-side, network entity (NW)-side, and / or bilateral AI / ML models in NWDAF and / or gNB. Training or retraining may be triggered by triggering MDT to collect physical layer measurements, and the UE may provide the physical layer measurements to the NWDAF or gNB.

[0033] Figure 3 An example of a flow chart for triggering AI / ML training or retraining using MDT according to various exemplary embodiments is shown. The process may include, for example, at 310, the gNB may send a request to the NWDAF to train or retrain one or more ML models for a feature or feature group for enabling AI / ML. At 320, the training or retraining request may trigger a request to download the model to the NWDAF from a provider (such as a UE, a gNB, a location management function (LMF), an operations and management (OAM) node, a core network (CN) node, and / or an over-the-top (OTT) server). At 330, the NWDAF may provide a confirmation to the provider providing the model(s) upon receiving the model(s), and may determine whether the NWDAF has the required amount of resources or has collected a pre-configured amount of data to perform training on the downloaded model(s). If the provider is not a gNB, the provider may forward the confirmation of receiving the model(s) from the NWDAF to the gNB.

[0034] At 340, if the NWDAF determines that the NWDAF has a predetermined amount of resources or has collected a predetermined amount of data to perform training or retraining, the NWDAF may train or retrain the (multiple) downloaded models. The predetermined amount of resources may be, for example, the number of CPUs / GPUs, the amount of memory, signaling / interface bandwidth, etc. At 350, if the NWDAF may not have a predetermined amount of resources or may not have collected a predetermined amount of data to perform training or retraining, the NWDAF may request the MaS to provide additional training data for training or retraining the (multiple) downloaded AI / ML models. The process may be repeated until the NWDAF determines that the NWDAF has a predetermined amount of resources or has collected a predetermined amount of data to perform training or retraining at process 340.

[0035] Various exemplary embodiments may provide one or more processes for training or retraining a model on the UE side, UE portion, NW side, and / or NW portion in a NW entity. One or more processes may include, for example, a NW entity (such as NWDAF) may download an AI / ML model and may start training the model based on the configuration of various resources (such as the availability of computing resources and the memory capacity for loading models and data). The NWDAF may assess data collection availability and adequacy, such as whether a predetermined amount of data has been collected, and may determine whether required additional auxiliary information from different NW entities and / or UEs may be collected and added to the data collection for training or retraining.

[0036] (Multiple) AI / ML models can be trained or retrained in parallel or sequentially using different provided hyperparameters, and (multiple) AI / ML models can be validated to select a model from candidate models based on the hyperparameters. NWDAF can then forward the trained models and data to different network entities for future use.

[0037] Certain exemplary embodiments may provide an initial training phase in which the NWDAF may train an untrained model that has not been trained when the NWDAF has both the required computing resources and data. The gNB may initialize the collection of training models or models for AI / ML-enabled features and / or functions. Initializing the training model may trigger a request to the NWDAF. The initialization request may be triggered by the UE, LMF, or gNB. The NWDAF may download (multiple) target models from the gNB, CN node, OAM node, LMF, and / or OTT server. The NWDAF may not be permanently stored. Upon receiving the (multiple) downloaded models, the NWDAF may confirm receipt of the (multiple) models. The NWDAF may assess or determine whether data collection is required, and if so, may request training data from the MaS. The NWDAF may then perform AI / ML training on the (multiple) models.

[0038] Some exemplary embodiments may provide a retraining phase. The gNB, UE and / or LMF may monitor the performance of one or more models. During the monitoring, the gNB, UE and / or LMF may evaluate the performance of one or more models. For example, the evaluation of the model based on the monitoring may be decided at the gNB. If one or more key performance indicators (KPIs) of the performance evaluation may not be considered satisfactory, the gNB may request retraining or updating of one or more models. Retraining or updating of one or more models will be performed by providing one or more models to the NWDAF and may trigger a data collection procedure in the NWDAF.

[0039] In addition, or as an alternative, the gNB or network entity with ML training capabilities may perform a relatively reduced process of updating and / or fine-tuning the initially trained model received from the NWDAF. The data collection process in the NWDAF may still be triggered, and data may also be collected in the gNB or network entity with ML training capabilities.

[0040] Certain exemplary embodiments may reduce signaling and resource overhead over the NR air interface by, for example, the gNB forwarding data for fine-tuning to the NWDAF. Alternatively, in some exemplary embodiments, the NWDAF may not need to perform updates and / or fine-tuning. The reduction in signaling and resource overhead over the NR air interface may reduce the delay in the new model being available at the gNB and / or UE.

[0041] After one or more models have been updated or fine-tuned by the gNB or an entity with AI / ML training capabilities, the gNB may deliver the one or more models to the NWDAF to be available for subsequent operations, such as additional fine-tuning and adjustment, federation, knowledge distillation, etc. For UE-side or UE-part models, data collection may be triggered by the UE, where the UE may request from the gNB the type of measurement data that the UE may report to the gNB for model retraining.

[0042] Various exemplary embodiments may provide that the MaS may create an MDT activation session involving the UDM, AMF, and / or gNB. For example, the gNB may configure the MDT session to the UE. After establishing the RRC connection, the UE may log measurement reports and may send measurement reports to the gNB. The gNB forwards the measurement logs to the Tracking Collection Entity (TCE). The TCE may forward the measurement logs to the NWDAF. The NWDAF may perform training and validation on the collected data. If the training entity (such as a gNB, LMF, etc.) is different from the NWDAF, the training and validation may be performed in the gNB and / or LMF. After the training or retraining is completed, one or more models may be delivered to the gNB, UE, LMF, CN, and / or OTT server for evaluation or storage.

[0043] Figure 4A and Figure 4B A signaling diagram is shown for triggering one or more processes for training or retraining a model, according to certain exemplary embodiments. Figure 4A and Figure 4B The network configuration in the example of FIG4 may include MaS 401, UDM 402, AMF 403, UE 404, a base station or a network node (such as gNB 405), TCE 406, and NWDAF 407. Figure 4A As shown in , the process for training or retraining a model may include that at 410, the gNB 405 may initialize the ML model training process, and at 411, the gNB 405 may request a previously trained model from the NWDAF 407. At 412, the NWDAF 407 may download the model to the gNB 405, and at 413, the gNB 405 may send a request for the trained model to the NWDAF 407.

[0044] At 414, NWDAF 407 may send a training data collection request to MaS 401. At 415, gNB 405 may perform a performance evaluation of the model and may determine that the model may be underperforming. At 416, gNB 405 may send a retraining request for the model to NWDAF 407, and at 417, NWDAF 407 may send a request for training data collection for the model to MaS. At 418, MaS 401 may send an MDT activation to UDM 402, and at 419, UDM 402 may send an MDT activation to AMF 403. At 420, AMF 403 may send an MDT activation to gNB 405, and at 421, gNB 405 may perform RRC reconfiguration with UE 404. At 422, UE 404 may complete RRC configuration with gNB 405.

[0045] like Figure 4B As shown in , the following operations 423-425 may be performed in a loop. At 423, UE 404 may send a measurement report to gNB 405. At 424, gNB 405 may send a measurement report to TCE 406, and at 425, TCE 406 may send a measurement report to NWDAF 407. At 426, NWDAF 407 may perform training and validation of the model.

[0046] The following operations 427-429 may be performed as sub-operations between NWDAF 407 and gNB 405. At 427, NWDAF 407 may send a notification of the completion of training of the model to gNB 405, and at 428, NWDAF 407 may deliver the trained model to gNB 405. At 429, gNB 405 may retrain or fine-tune and run the trained model from NWDAF 407 based on UE data from measurement reports, and at 429A, gNB 405 may send the newly trained model to NWDAF 407 for storage or distribution to other UEs.

[0047] At 430, NWDAF 407 may optionally provide initial delivery of the model to gNB 405, and at 431, UE 404 may send measurement reports to gNB 405, which may occur cyclically, repeatedly, or at recurring intervals. At 432, gNB 405 may train and validate the model based on the data in the measurement reports. At 433, gNB 405 may deliver the model to NWDAF 407, and at 434, gNB 405 may execute the model.

[0048] Some exemplary embodiments may provide an alternative where, instead of the TCE 406 forwarding the measurement reports to the NWDAF 407, the gNB 405 may forward the measurement reports directly to the NWDAF 407. In some cases, this may result in faster data collection for training purposes.

[0049] Figure 5 An example of a signaling diagram of one or more procedures for direct data forwarding is shown in accordance with certain exemplary embodiments. Figure 5 The network configuration in the example may include MaS 501, UDM 502, AMF 503, UE 504, a base station or network node (such as gNB 505), and NWDAF 506. The process for training or retraining a model may include: at 510, gNB 505 may initialize the ML model training process; and at 511, gNB 505 may request a previously trained model from NWDAF 506. At 512, NWDAF 506 may download the model to gNB 505; and at 513, gNB 505 may send a request for a trained model to NWDAF 506.

[0050] At 514, NWDAF 506 may send a model training request to MaS 501; and at 515, gNB 505 may perform a performance evaluation of the model and may determine that the model may be underperforming. At 516, gNB 505 may send a retraining request for the model to NWDAF 506, and at 517, NWDAF 506 may send a request for training data collection for the model to MaS 501. At 518, MaS 501 may send MDT activation to UDM 502, and at 519, UDM 502 may send MDT activation to AMF 503. At 520, AMF 503 may send MDT activation to gNB 505, and at 521, gNB 505 may perform RRC reconfiguration with UE 504. At 522, UE 504 may complete RRC configuration with gNB 505, and at 523, UE 504 may send a measurement report to gNB 505. At 524, gNB 505 may send the measurement report to NWDAF 506. At 525, NWDAF 506 may perform training and validation of the model.

[0051] At 526, NWDAF 506 may send a notification to gNB 505 that training of the model is complete, and at 527, NWDAF 506 may deliver the trained model to gNB 505. At 528, gNB 505 may apply and / or execute the trained model.

[0052] Figure 6 An example flow chart of a method according to some exemplary embodiments is shown. In some exemplary embodiments, Figure 6 The method may be performed by a network node, a network element, or a group of multiple network nodes or elements in a 3GPP system (such as LTE or 5G-NR). For example, in some exemplary embodiments, Figure 6 The method can be similar to Figure 8 It is performed by the NWDAF of the device 810 shown in the figure.

[0053] According to various exemplary embodiments, Figure 6 The method may include obtaining, at 610, from a network entity at least one machine learning model for training or retraining based on measurement data of the network. At 620, the method may include determining that data collection is required before training or retraining the at least one machine learning model, and receiving one or more measurement reports including at least one data set from the data collection from the network function. The method may also include determining, at 630, whether additional auxiliary information from one or more network devices requires training or retraining. At 640, the method may include training or retraining the at least one machine learning model based on all collected data sets, including at least one data set from the data collection.

[0054] Some exemplary embodiments may provide that the method further comprises receiving one or more additional machine learning models to be trained or retrained in the future. The obtained machine learning model for training or retraining may be selected from a plurality of machine learning models accessible to the device 810. The method may include sending a confirmation to a network entity from which the device obtained at least one machine learning model for training or retraining. The method may also verify the trained model based on one or more measurement reports.

[0055] Certain exemplary embodiments may provide that the method further comprises providing the trained model to the network device or the user equipment. The method may further comprise receiving a request from the network entity to retrain at least one machine learning model, sending a request to the network function for training data for at least one machine learning model to be retrained, and retraining the machine learning model based on the requested training data.

[0056] Figure 7 An example flow chart of a method according to some exemplary embodiments is shown. In some exemplary embodiments, Figure 7 The method may be performed by a network node, a network element, or a group of multiple network nodes or elements in a 3GPP system (such as LTE or 5G-NR). For example, in some exemplary embodiments, Figure 7 The method can be similar to Figure 8 The base station or gNB of the device 820 shown in the figure is executed.

[0057] According to various exemplary embodiments, Figure 7 The method may include determining, at 710, that at least one machine learning model may need to be updated, and at 720, the method may include sending a request to a network entity to train or retrain at least one machine learning model. At 730, the method may also include providing at least one measurement report indicating performance of the user equipment to the network entity, and at 740, receiving a trained or retrained model to be executed on the device 820.

[0058] Certain exemplary embodiments may provide that the method may further include receiving one or more additional machine learning models to be trained or retrained in the future. The obtained machine learning model for training or retraining may be selected from a plurality of machine learning models accessible to the device 820. The method may further include validating the trained machine learning model based on one or more measurement reports and executing the trained machine learning model. The method may further include providing a request to retrain at least one machine learning model to a network entity, receiving one or more measurement reports including at least one data set for retraining at least one machine learning model, and retraining the machine learning model based on the one or more measurement reports.

[0059] Figure 8 A set of devices 810 and 820 according to various exemplary embodiments are shown. In various exemplary embodiments, the device 810 may be a network, a network entity, an element of a core network, or an element in a communication network or an element associated with such a network. For example, a NWDAF according to various exemplary embodiments as discussed above may be an example of the device 810. It should be noted that a person of ordinary skill in the art will understand that the device 810 may include Figure 8 Components or features not shown in the figure. In addition, the device 820 can be an element of a network, a network entity, a core network, or an element in a communication network or an element associated with such a network, such as a base station, a NE, or a gNB. For example, a gNB according to the various exemplary embodiments discussed above can be an example of the device 820. It should be noted that a person of ordinary skill in the art will understand that the device 820 may include Figure 8 Components or features not shown.

[0060] According to various exemplary embodiments, the apparatus 810 may include at least one processor and at least one memory, such as Figure 8As shown in . The memory may store instructions that, when executed by the processor, cause the device 810 to obtain at least one model for training or retraining based on network measurement data. The device 810 may also be caused to determine that data collection is required before training or retraining at least one model, and receive one or more measurement reports including at least one data set from the data collection. The device 810 may also be caused to train or retrain at least one model based on all collected data sets, including at least one data set from the data collection.

[0061] According to various exemplary embodiments, the apparatus 820 may include at least one processor and at least one memory, such as Figure 8 As shown in . The memory may store instructions that, when executed by the processor, cause the device 820 to determine that at least one model performs poorly and send a request to a network entity to train or retrain at least one model. The device 820 may also provide at least one measurement report indicating the performance of the user equipment to the network entity and receive a training or retraining model to be performed on the device 820.

[0062] The various exemplary embodiments described above may provide several technical improvements, enhancements and / or advantages. For example, some exemplary embodiments may provide advantages and / or improvements to NWDAF, gNB and / or network to traditional MDT configuration procedures by, for example, implementing NWDAF and triggering data collection of RAN measurements through MDT to train or retrain an AI / ML model offline.

[0063] In some exemplary embodiments, the apparatus 810 and / or 820 may include one or more processors, one or more computer-readable storage media (e.g., memory, storage, etc.), one or more radio access components (e.g., modem, transceiver, etc.), and / or a user interface. In some exemplary embodiments, the apparatus 810 and / or 820 may be configured to operate using one or more radio access technologies, such as GSM, LTE, LTE-A, NR, 5G, WLAN, WiFi, NB-IoT, Bluetooth, NFC, MulteFire, and / or any other radio access technology.

[0064] like Figure 8As shown in the example of , the device 810 and / or 820 may include or be coupled to a processor 812 and 822, respectively, for processing information and executing instructions or operations. The processors 812 and 822 may be any type of general or special purpose processor. In fact, as examples, the processors 812 and 822 may include one or more of the following: a general purpose computer, a special purpose computer, a microprocessor, a digital signal processor (DSP), a field programmable gate array (FPGA), an application specific integrated circuit (ASIC), and a processor based on a multi-core processor architecture. Although in Figure 8 A single processor 812 (and 822) is shown for each of the devices 810 and / or 820, but multiple processors may be utilized according to other example embodiments. For example, it should be understood that in some example embodiments, the devices 810 and / or 820 may include two or more processors that may form a multiprocessor system (e.g., in which case, the processors 812 and 822 may represent a multiprocessor) that may support multiprocessing. According to some example embodiments, the multiprocessor system may be tightly coupled or loosely coupled, for example to form a computer cluster.

[0065] Processors 812 and 822 may perform functions associated with the operation of devices 810 and / or 820, respectively, including, as some examples, precoding of antenna gain / phase parameters, encoding and decoding of individual bits forming communication messages, formatting of information, and overall control of devices 810 and / or 820, including Figure 3-7 The process is shown in the figure.

[0066] Devices 810 and / or 820 may also include or be coupled to memory 814 and / or 824 (internal or external), respectively, which may be coupled to processors 812 and 822, respectively, for storing information and instructions that may be executed by processors 812 and 822. Memory 814 (and memory 824) may be one or more memories and of any type suitable for the local application environment, and may be implemented using any suitable volatile or non-volatile data storage technology, such as semiconductor-based memory devices, magnetic memory devices and systems, optical memory devices and systems, fixed memory, and / or removable memory. For example, memory 814 (and memory 824) may be any combination of random access memory (RAM), read-only memory (ROM), static memory (such as a magnetic disk or optical disk), hard disk drive (HDD), or any other type of non-transitory machine or computer-readable medium. The instructions stored in memory 814 and memory 824 may include program instructions or computer program code that, when executed by processors 812 and 822 , enable apparatus 810 and / or 820 to perform tasks as described herein.

[0067] In certain example embodiments, the apparatus 810 and / or 820 may also include or be coupled to a (internal or external) drive or port configured to accept and read an external computer-readable storage medium, such as an optical disk, a USB drive, a flash drive, or any other storage medium. For example, the external computer-readable storage medium may store a computer program or software for execution by the processors 812 and 822 and / or the apparatus 810 and / or 820 to perform Figure 3-7 Any of the methods illustrated in .

[0068] In some example embodiments, an apparatus (e.g., apparatus 810 and / or apparatus 820) may include components for performing a method, process, or any variant discussed herein. Examples of components may include one or more processors, memories, controllers, transmitters, receivers, and / or computer program code for causing the execution of operations.

[0069] Various exemplary embodiments may be directed to an apparatus, such as apparatus 810, comprising means for obtaining at least one model for training or retraining based on measurement data of a network. The apparatus may also include means for determining that data collection is required before training or retraining at least one model and receiving one or more measurement reports including at least one data set from the data collection. The apparatus may also include means for training or retraining at least one model based on all collected data sets, including at least one data set from the data collection.

[0070] Certain exemplary embodiments may be directed to an apparatus, such as apparatus 820, comprising means for determining that at least one model requires updating and means for sending a request to a network entity to train or retrain at least one model. The apparatus may also include means for providing at least one measurement report indicating performance of a user equipment to the network entity, and means for receiving a training or retraining model to be performed on the apparatus.

[0071] In some exemplary embodiments, the device 810 may also include or be coupled to one or more antennas 815 for receiving downlink signals and transmitting from the device 810 via an uplink. The devices 810 and / or 820 may also include transceivers 816 and 826, respectively, which are configured to send and receive information. The transceivers 816 and 826 may also include a radio interface, which may correspond to one or more of a variety of radio access technologies, including GSM, LTE, LTE-A, 5G, NR, WLAN, NB-IoT, Bluetooth, BT-LE, NFC, RFID, UWB, etc. The radio interface may include other components, such as filters, converters (e.g., digital-to-analog converters, etc.), symbol demappers, signal shaping components, inverse fast Fourier transform (IFFT) modules, etc., to process symbols carried by the downlink or uplink, such as OFDMA symbols.

[0072] For example, transceivers 816 and 826 may be configured to modulate information onto a carrier waveform for transmission, and demodulate received information for further processing by other elements of device 810 and / or 820, respectively. In other example embodiments, transceivers 816 and 826 may be capable of directly sending and receiving signals or data. Additionally or alternatively, in some example embodiments, device 810 and / or 820 may include input and / or output devices (I / O devices). In certain example embodiments, device 810 and / or 820 may also include a user interface, such as a graphical user interface or a touch screen.

[0073] In certain example embodiments, memory 814 and memory 824 store software modules that provide functionality when executed by processors 812 and 822, respectively. The modules may include, for example, an operating system that provides operating system functionality for device 810 and / or 820. The memory may also store one or more functional modules, such as applications or programs, to provide additional functionality to device 810 and / or 820. The components of device 810 and / or 820 may be implemented in hardware, or in any suitable combination of hardware and software. According to certain example embodiments, device 810 may optionally be configured to communicate with device 820 via a wireless or wired communication link 830 according to any radio access technology, such as NR.

[0074] According to some example embodiments, the processors 812 and 822 and the memories 814 and 824 may be included in a processing circuit or a control circuit or may form a part of a processing circuit or a control circuit. In addition, in some example embodiments, the transceivers 816 and 826 may be included in a transceiver circuit or may form a part of a transceiver circuit.

[0075] As used herein, the term "circuit" may refer to any portion of a hardware circuit implementation (e.g., analog and / or digital circuits), a combination of hardware circuits and software, a combination of analog and / or digital hardware circuits and software / firmware, a hardware processor (including a digital signal processor) with software that works together to enable a device (e.g., device 810 and / or 820) to perform various functions, and / or (multiple) hardware circuits and / or (multiple) processors or portions thereof that operate using software, but the software may not be present when not required for operation. As another example, as used herein, the term "circuit" may also cover an implementation of a hardware circuit or processor or multiple processors alone, or a portion of a hardware circuit or processor, and accompanying software and / or firmware. The term circuit may also cover, for example, a baseband integrated circuit in a server, cellular network node or device, or other computing or network device.

[0076] The computer program product may include one or more computer executable components that are configured to perform some example embodiments when the program is run. The one or more computer executable components may be at least one software code or part thereof. Modifications and configurations required to implement the functionality of some example embodiments may be performed as (multiple) routines, which may be implemented as (multiple) added or updated software routines. (Multiple) software routines may be downloaded to a device.

[0077] As an example, software or computer program code or parts thereof may be in source code form, object code form or some intermediate form, and may be stored in some carrier, distribution medium or computer readable medium, which may be any entity or device capable of carrying the program. For example, such carriers may include recording media, computer memory, read-only memory, optoelectronic and / or electrical carrier signals, telecommunication signals and software distribution packages. Depending on the processing power required, the computer program may be executed in a single electronic digital computer, or it may be distributed among multiple computers. The computer readable medium or computer readable storage medium may be a non-transitory medium.

[0078] In other example embodiments, the functionality may be performed by hardware or circuitry included in a device (e.g., device 810 and / or 820), such as by using an application specific integrated circuit (ASIC), a programmable gate array (PGA), a field programmable gate array (FPGA), or any other combination of hardware and software. In yet another example embodiment, the functionality may be implemented as a signal, i.e., a non-tangible component that may be carried by an electromagnetic signal downloaded from the Internet or other network.

[0079] According to certain example embodiments, an apparatus (such as a node, a device or a corresponding component) may be configured as a circuit, a computer or a microprocessor (such as a single-chip computer element) or as a chipset, including at least a memory for providing storage capacity for arithmetic operations and an operation processor for performing arithmetic operations.

[0080] The features, structures, or characteristics of the example embodiments described throughout this specification may be combined in any suitable manner in one or more example embodiments. For example, the use of the phrases "certain embodiments," "example embodiments," "some embodiments," or other similar language throughout this specification refers to the fact that a particular feature, structure, or characteristic described in conjunction with an embodiment may be included in at least one embodiment. Therefore, the appearance of the phrases "in certain embodiments," "example embodiments," "in some embodiments," "in other embodiments," or other similar language throughout this specification does not necessarily refer to the same set of embodiments, and the described features, structures, or characteristics may be combined in any suitable manner in one or more example embodiments. In addition, the terms "cell," "node," "gNB," or other similar language may be used interchangeably throughout this specification.

[0081] As used herein, “at least one of: ” and “at least one of ” and similar expressions mean at least any one of the elements, or at least any two or more of the elements, or at least all of the elements, where the list of two or more elements is connected by “and” or “or”.

[0082] A person of ordinary skill in the art will readily appreciate that the present disclosure as discussed above may be practiced using processes in a different order and / or using hardware elements in a configuration different from that disclosed. Therefore, although the present disclosure has been described based on these example embodiments, certain modifications, variations, and alternative constructions will be apparent to those skilled in the art while remaining within the spirit and scope of the example embodiments. Although the above embodiments relate to 5G NR and LTE technologies, the above embodiments may also be applicable to any other current or future 3GPP technologies, such as LTE Advanced and / or fourth generation (4G) technologies.

[0083] Partial Glossary :

[0084] 3GPP Third Generation Partnership Project

[0085] 5G Fifth Generation

[0086] ACK

[0087] AMF Access and Mobility Function

[0088] CN Core Network

[0089] DL Downlink

[0090] EMBB Enhanced Mobile Broadband

[0091] gNB 5G or Next Generation NodeB

[0092] LMF Location Management Function

[0093] LTE Long Term Evolution

[0094] MaS Management Services

[0095] NR New Radio

[0096] NWDAF network data analysis function

[0097] OAM Operation and Management

[0098] OTT Over The Top

[0099] RAN Radio Access Network

[0100] RedCap Reduction Capacity

[0101] RRC Radio Resource Control

[0102] TCE Tracking Collection Entity

[0103] UDM Unified Data Management

[0104] UE User Equipment

[0105] UL Uplink

[0106] URLLC ultra-reliable low-latency communication

Claims

1. A device for communication, comprising: at least one processor; as well as at least one memory storing instructions that, when executed by the at least one processor, cause the apparatus to at least: obtaining from a network entity at least one machine learning model for training or retraining based on measurement data of the network; determining that data collection is required prior to training or retraining the at least one machine learning model, and receiving one or more measurement reports including at least one data set from the data collection from a network function; determining whether additional auxiliary information from one or more network devices requires training or retraining; as well as The at least one machine learning model is trained or retrained based on all collected data sets including the at least one data set from the data collection.

2. The apparatus of claim 1 , wherein the at least one memory and the instructions, when executed by the at least one processor, further cause the apparatus to at least: Receiving one or more additional models to be trained or retrained in the future, wherein the obtained machine learning model for training or retraining is selected from a plurality of machine learning models accessible to the device.

3. The apparatus of claim 1 , wherein the at least one memory and the instructions, when executed by the at least one processor, further cause the apparatus to at least: Sending a confirmation to the network entity from which the device obtained the at least one machine learning model for training or retraining.

4. The apparatus of claim 1 , wherein the at least one memory and the instructions, when executed by the at least one processor, further cause the apparatus to at least: The trained model is validated based on the one or more measurement reports.

5. The apparatus of claim 1 , wherein the at least one memory and the instructions, when executed by the at least one processor, further cause the apparatus to at least: The trained model is provided to a network device or a user device.

6. The apparatus of any one of claims 1-5, wherein the at least one memory and the instructions, when executed by the at least one processor, further cause the apparatus to at least: receiving, from the network entity, a request to retrain the at least one machine learning model; sending a request to the network function for training data for the at least one machine learning model to be retrained; as well as Retrain the at least one machine learning model based on the requested training data.

7. An apparatus for communication, comprising: at least one processor; as well as at least one memory storing instructions that, when executed by the at least one processor, cause the apparatus to at least: Determine that at least one machine learning model needs to be updated; sending a request to a network entity to train or retrain the at least one machine learning model; providing at least one measurement report to the network entity; as well as A trained or retrained machine learning model is received from the network entity to be executed on the device.

8. The apparatus of claim 7, wherein the at least one memory and the instructions, when executed by the at least one processor, further cause the apparatus to at least: Receiving one or more additional machine learning models to be trained or retrained in the future, wherein the obtained machine learning models for training or retraining are selected from a plurality of machine learning models accessible to the device.

9. The apparatus of claim 7, wherein the at least one memory and the instructions, when executed by the at least one processor, further cause the apparatus to at least: validating the trained machine learning model based on the one or more measurement reports; and Execute the trained machine learning model.

10. The apparatus of any one of claims 7-9, wherein the at least one memory and the instructions, when executed by the at least one processor, further cause the apparatus to at least: providing, to the network entity, a request to retrain the at least one machine learning model; receiving one or more measurement reports comprising at least one data set for retraining the at least one machine learning model; and The at least one machine learning model is retrained based on the one or more measurement reports.

11. A method for communication, comprising: obtaining, by the device from a network entity, at least one machine learning model for training or retraining based on measurement data of the network; determining that data collection is required prior to training or retraining the at least one machine learning model, and receiving one or more measurement reports including at least one data set from the data collection from a network function; determining whether additional auxiliary information from one or more network devices requires training or retraining; as well as The at least one machine learning model is trained or retrained based on all collected data sets including the at least one data set from the data collection.

12. The method according to claim 11, further comprising: Receiving one or more additional models to be trained or retrained in the future, wherein the obtained machine learning model for training or retraining is selected from a plurality of machine learning models accessible to the device.

13. The method according to claim 11, further comprising: Sending a confirmation to the network entity from which the device obtained the at least one machine learning model for training or retraining.

14. The method according to claim 11, further comprising: The trained model is validated based on the one or more measurement reports.

15. The method according to claim 11, further comprising: The trained model is provided to a network device or a user device.

16. The method according to any one of claims 11 to 15, further comprising: receiving, from the network entity, a request to retrain the at least one machine learning model; sending a request to the network function for training data for the at least one machine learning model to be retrained; as well as Retrain the at least one machine learning model based on the requested training data.

Images