UE-initiated model updates for bilateral AI / ML models

Abstract

A method includes obtaining, at a terminal assigned to a network, a configuration having parameters that enable the terminal to initiate an uplink, UL, transmission relating to an update at the terminal at least for a terminal portion of a bilateral model used at the terminal, wherein the bilateral model is used for joint model reasoning at a terminal side and a network side; determining that update of the bilateral model is required; initiating a first UL transmission based on the obtained configuration; sending the initiated first UL transmission, the initiated first UL transmission indicating the required update; receiving a response, the response comprising: update information related to the required update; and establishing an update of the terminal portion of the bilateral model based on the received response.

Description

Technical Field

[0001] The present disclosure relates to a method and apparatus for user equipment (UE) initiated model update for a bilateral artificial intelligence / machine learning (AI / MI) model. Background Art

[0002] The following description of the background technology may include insights, discoveries, understandings or disclosures or associations to at least some examples of embodiments of the present disclosure, together with disclosures not known in the relevant prior art, but provided by the present disclosure. Some of such contributions of the present disclosure may be specifically pointed out below, while other such contributions of the present disclosure will be apparent from the relevant context.

[0003] In recent years, the continuous expansion of communication networks, such as wire-based communication networks, such as integrated services digital network (ISDN), digital subscriber line (DSL), or wireless communication networks, such as cdma2000 (code division multiple access) system, cellular third generation (3G), such as universal mobile telecommunications system (UMTS), fourth generation (4G) communication networks or enhanced communication networks based on, for example, long term evolution (LTE) or advanced long term evolution (LTE-A), fifth generation (5G) communication networks, cellular second generation (2G) communication networks, such as global system for mobile communications (GSM), general packet radio system (GPRS), enhanced data rates for global evolution (EDGE), or other wireless communication systems (such as wireless local area network (WLAN), Bluetooth or world wide interoperability for microwave access (WiMAX)) have appeared all over the world. Various organizations such as European Telecommunications Standards Institute (ETSI), 3rd Generation Partnership Project (3GPP), Telecommunication and Internet Converged Services and Advanced Networking Protocol (TISPAN), International Telecommunication Union (ITU), 3rd Generation Partnership Project 2 (3GPP2), Internet Engineering Task Force (IETF), IEEE (Institute of Electrical and Electronics Engineers), WiMAX Forum, etc. are developing standards or specifications for telecommunication networks and access environments.

[0004] In such context, 3GPP document RP-213599, Release 18 Study Item (SI) on Artificial Intelligence (AI) / Machine Learning (ML) for New Radio (NR) Air Interface, aims to explore the benefits of enhancing the air interface with features that enable AI / ML-based algorithms for enhanced performance and / or reduced complexity / overhead. The goal of this SI is to lay the foundation for future air interface use cases that leverage AI / ML techniques. The initial set of use cases to be covered include channel state information (CSI) feedback enhancement (e.g., overhead reduction, improved accuracy, prediction), beam management (e.g., beam prediction in the time and / or spatial domain for overhead and latency reduction, improved beam selection accuracy), positioning accuracy enhancement. For those use cases, the benefits should be evaluated (using developed methodologies and defined key performance indicators (KPIs)) and the potential impact on the specification should be evaluated, including PHY layer aspects, protocol aspects.

[0005] One of the key expected outcomes of the SI is that the AI / ML approaches need to be diverse enough to support the various requirements regarding the level of gNB-UE collaboration.

[0006] Additionally, it is important to note that other use cases may also be addressed during the WI phase of “AI / ML for air interfaces”. Starting from Release 18, companies will most likely propose a wide variety of use cases and applications for ML in gNB and UE. The goal is to explore the benefits of enhancing the air interface with features that enable AI / ML-based algorithms for enhanced performance and / or reduced complexity / overhead. The enhanced performance depends on the use case considered and may, for example, improve throughput, robustness, accuracy or reliability. The goal is that enough use cases will be considered to enable the identification of a common AI / ML framework, including functional requirements for the AI / ML architecture, which can be used for subsequent projects. The study should also identify areas where AI / ML can improve the performance of air interface functions. The regulatory impact will be evaluated in order to fully understand what is needed to enable AI / ML technologies for air interfaces.

[0007] In this regard, for ease of understanding, the bilateral model will be outlined in more detail below.

[0008] Regarding the bilateral model, it should be noted that in the RAN1#109-e meeting, RAN1 made working assumptions on AI / ML model terminology.

[0009] The working assumption is that the following will be included in the working list of terms to be used for RAN1 AI / ML air interface SI discussions (see Table 1). The descriptions of the terms may be further refined as the study progresses, and new terms may be added as the study progresses.

[0010] Table 1: Terminology working list

[0011]

[0012]

[0013] The bilateral model mainly considers CSI compression, in which the following agreements (i) to (iii) were reached at the RAN1#109-e and RAN1#110 meetings.

[0014] (i) That is, for the evaluation of the AI / ML-based CSI compression sub-use case, the bilateral model is considered as the starting point, including an AI / ML-based CSI generation part to generate CSI feedback information and an AI / ML-based CSI reconstruction part, which is used to reconstruct the CSI from the received CSI feedback information.

[0015] At least for inference, the CSI generation part is located on the UE side and the CSI reconstruction part is located on the gNB side.

[0016] (ii) In addition, spatial-frequency domain CSI compression using a bilateral AI model is selected as a representative sub-use case.

[0017] It should be noted that the study of other sub-use cases is not excluded.

[0018] It should also be noted that all pre-processing / post-processing, quantization / dequantization are within the scope of the sub-use case.

[0019] (ii) In addition, in CSI compression using bilateral model use case, the following AI / ML model training collaboration will be further studied:

[0020] Type 1: A bilateral model is jointly trained on a single side / entity (e.g., UE side or network side).

[0021] Type 2: Jointly train bilateral models on the network side and the UE side, respectively.

[0022] Type 3: Separate training at the network side and the UE side, where the UE side CSI generation part and the network side CSI reconstruction part are trained by the UE side and the network side respectively.

[0023] It should be noted that joint training means that the generative model and the reconstruction model should be trained in the same cycle with forward and backward propagation. Joint training can be performed both in a single node or across multiple nodes (e.g., by exchanging gradients between nodes).

[0024] It should also be noted that separate training includes sequential training starting from UE side training, or sequential training starting from NW side training, or parallel training at UE and NW.

[0025] Other types of collaboration are not excluded.

[0026] Overall, bilateral models pose certain challenges and require careful consideration of how such models can be used in the NR framework. Therefore, in view of this, it is necessary to address at least some of these issues associated with such bilateral models.

[0027] Therefore, the purpose of this specification is to improve the prior art.

[0028] The abbreviations used in this specification apply to:

[0029] 2G Second Generation

[0030] 3G Third Generation

[0031] 3GPP Third Generation Partnership Project

[0032] 3GGP2 3rd Generation Partnership Project 2

[0033] 4G Fourth Generation

[0034] 5G Fifth Generation

[0035] 6G Sixth Generation

[0036] AI

[0037] AN Access Node

[0038] AP Access Point

[0039] BS Base Station

[0040] CDMA Code Division Multiple Access

[0041] CNN Convolutional Neural Network

[0042] CSI Channel State Information

[0043] DCI Downlink Control Information

[0044] DL Downlink

[0045] DSL Digital Subscriber Line

[0046] EDGE Enhanced Data Rates for Global Evolution

[0047] EEPROM Electrically Erasable Programmable Read Only Memory

[0048] eNB Evolved Node B

[0049] ETSI European Telecommunications Standards Institute

[0050] gNB Next Generation Node B

[0051] GPRS General Packet Radio System

[0052] GSM Global System for Mobile Communications

[0053] IEEE Institute of Electrical and Electronics Engineers

[0054] ISDN Integrated Services Digital Network

[0055] ITU International Telecommunication Union

[0056] KPI Key Performance Indicator

[0057] LTE Long Term Evolution

[0058] LTE-A Long Term Evolution Advanced

[0059] MAC CE Media Access Control Unit

[0060] MANETs Mobile Ad Hoc Networks

[0061] MIMO Multiple Input Multiple Output

[0062] ML Machine Learning

[0063] NR New Radio

[0064] NB Node B

[0065] NN Neural Network

[0066] RAM Random Access Memory

[0067] RAN Radio Access Network

[0068] RB residual block

[0069] ROM Read Only Memory

[0070] RS reference signal

[0071] Rx Receiver

[0072] SSB Synchronous Signal Block

[0073] TISPAN Telecom and Internet Convergence Services and Advanced Network Protocols

[0074] TRP Transmission Point

[0075] Tx Transmitter

[0076] UE User Equipment

[0077] UL Uplink

[0078] UMTS Universal Mobile Telecommunications System

[0079] UWB Ultra Wideband

[0080] WCDMA Wideband Code Division Multiple Access

[0081] WiMAX Microwave Access Global Interoperability

[0082] WLAN Wireless Local Area Network Summary of the invention

[0083] The purpose of various examples of embodiments of the present disclosure is to improve the prior art. Therefore, at least some examples of embodiments of the present disclosure are intended to solve at least part of the above-mentioned issues and / or problems and disadvantages.

[0084] Various aspects of examples of embodiments of the present disclosure are set out in the appended claims and relate to methods, apparatus and computer program products related to UE-initiated model updates for bilateral AI / ML models.

[0085] This object is achieved by the method, the device and the non-transitory storage medium specified in the appended claims. Advantageous further developments are set forth in the respective dependent claims.

[0086] According to any of the aspects mentioned in the appended claims, UE-initiated model updates for bilateral AI / ML models are enabled, thereby allowing at least part of the problems and disadvantages as identified / derivable above to be solved. Therefore, improvements are achieved by methods, apparatuses and computer program products that enable UE-initiated model updates for bilateral AI / ML models.

[0087] In more detail, the disclosure according to this specification allows ensuring that model reasoning is consistent between nodes involved in bilateral model reasoning.

[0088] Further advantages will become apparent from the following detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0089] Some embodiments of the present disclosure are described below by way of example only and with reference to the accompanying drawings, in which:

[0090] Figure 1 An example of training for a bilateral model is shown;

[0091] Figure 2 An example for ML model or dataset update on the UE side of a bilateral model is shown;

[0092] Figure 3 An example for ML model or dataset update on the network side of a bilateral model is shown;

[0093] Figure 4shows a message sequence for a bilateral ML model update scenario according to various example embodiments;

[0094] Figure 5 A flowchart showing steps corresponding to the method according to various examples of the embodiment is shown;

[0095] Figure 6 A flowchart showing steps corresponding to the method according to various examples of the embodiment is shown;

[0096] Figure 7 A block diagram showing an apparatus according to various examples of embodiments; and

[0097] Figure 8 Block diagrams of apparatuses according to various examples of embodiments are shown. DETAILED DESCRIPTION

[0098] Basically, in order to correctly establish and process communications between two or more endpoints (e.g., communication stations or elements or functions, such as terminal devices, user equipment (UE) or other communication network elements, databases, servers, hosts, etc.), one or more network elements or functions (e.g., virtualized network functions) may be involved, such as communication network control elements or functions (such as access points (APs), radio base stations (BSs), relay stations, eNBs, gNBs and other access network elements), and core network elements or functions (e.g., control nodes, support nodes, service nodes, gateways, user plane functions, access and mobility functions, etc.), which may belong to one communication network system or different communication network systems.

[0099] In the following, different exemplary embodiments will be described using a communication network architecture based on the 3GPP standard (such as 5G / NR) as an example of a communication network to which the embodiments can be applied, but the embodiments are not limited to such an architecture. It is obvious to those skilled in the art that these embodiments can also be applied to mobile communication principles (such as Wi-Fi, Worldwide Interoperability for Microwave Access (WiMAX), Personal Communications Service (PCS), Other types of communication networks, such as 4G and / or LTE (even 6G), are integrated with broadband code division multiple access (WCDMA), systems using ultra-wideband (UWB) technology, mobile ad hoc networks (MANET), wired access, etc.). In addition, without loss of generality, the description of some examples of the embodiments is related to mobile communication networks, but the principles of the present disclosure can be extended and applied to any other type of communication network, such as a wired communication network or a data center network.

[0100] The following examples and embodiments should be understood as illustrative examples only. Although this specification may refer to "one", "an" or "some" examples or embodiments in several places, this does not necessarily mean that each such reference is related to the same (multiple) examples or (multiple) embodiments, or that the feature applies only to a single example or embodiment. Individual features of different embodiments may also be combined to provide other embodiments. In addition, terms such as "including" and "comprising" should be understood as not limiting the embodiments to only the features mentioned; such examples and embodiments may also include features, structures, units, modules, etc. that have not been specifically mentioned.

[0101] The basic system architecture of a (telecommunication) communication network including a mobile communication system (in which some examples of the embodiments apply) may include the architecture of one or more communication networks, including (multiple) radio access network subsystems and (multiple) core networks. Such an architecture may include one or more communication network control elements or functions, access network elements, radio access network elements, access service network gateways or transceiver base stations, such as base stations (BS), access points (AP), NodeBs (NBs), eNBs or gNBs, distributed or centralized units (CUs), which control the corresponding coverage area or (multiple) cells, and one or more communication stations (such as communication elements or functions, such as user equipment (e.g., customer equipment), mobile devices or terminal devices (such as UEs), or another device with similar functions (such as modem chipsets, chips, modules, etc.), which may also be part of a station, element, function or application capable of communication, such as a UE, an element or function that can be used for a machine-to-machine communication architecture, or as a separate element attached to such an element, function or application capable of communication, etc.) utilizing its ability to communicate via one or more communication beams via one or more channels to send several types of data in multiple access domains. In addition, (core) network elements or network functions ((core) network control elements or network functions, (core) network management elements or network functions), such as gateway network elements / functions, mobility management entities, mobile switching centers, servers, databases, etc. may be included.

[0102] The general functionality and interconnection of the elements and functions (which also depends on the actual network type) are known to those skilled in the art and are described in the corresponding specifications, so that a detailed description thereof is omitted herein. However, it should be noted that several additional network elements and signaling links may be employed to communicate with elements, functions or applications, such as communication endpoints, communication network control elements (such as servers, gateways, radio network controllers) and other elements of the same or other communication networks in addition to those described in detail below.

[0103] The communication network architecture considered in the example of the embodiment may also be able to communicate with other networks, such as a public switched telephone network or the Internet. The communication network may also be able to support the use of cloud services for virtual network elements or their functions, where it should be noted that the virtual network portion of the telecommunications network may also be provided by non-cloud resources, such as an internal network, etc. It should be understood that the network elements and / or corresponding functions of the access system of the core network, etc. may be implemented by using any node, host, server, access node or entity, etc. suitable for such purpose. Typically, the network functions may be implemented as network elements on dedicated hardware, software instances running on dedicated hardware, or virtualized functions instantiated on an appropriate platform, such as a cloud infrastructure.

[0104] In addition, network elements (such as communication elements, such as UE, mobile devices, terminal devices), control elements or functions (such as access network elements, such as base stations (BS), eNB / gNB, radio network controllers), core network control elements or functions (such as gateway elements) or other network elements or functions (as described herein), (core) network management elements or functions and any other elements, functions or applications) can be implemented by software (e.g., by a computer program product for a computer) and / or by hardware. In order to perform its corresponding processing, the corresponding device, node, function or network element used may include several components, modules, units, components, etc. (not shown) required for control, processing and / or communication / signaling functions. Such components, modules, units and components may include, for example, one or more processors or processor units, including one or more processing parts for executing instructions and / or programs and / or for processing data, storage or memory units or components for storing instructions, programs and / or data, for use as a work area for a processor or processing part, etc. (e.g. ROM, RAM, EEPROM, etc.), input or interface components for inputting data and instructions by software (e.g. floppy disk, CD-ROM, EEPROM, etc.), user interfaces for providing monitoring and operation possibilities to the user (e.g. screen, keyboard, etc.), other interfaces or components for establishing links and / or connections under the control of the processor unit or part (e.g. wired and wireless interface components, radio interface components including, for example, antenna units, etc., components for forming a radio communication part, etc.), etc., wherein the corresponding components forming the interface (such as the radio communication part) may also be located at a remote site (e.g. a radio head or radio station, etc.). It should be noted that in the present specification, a processing part should not be regarded as representing only a physical part of one or more processors, but may also be regarded as a logical division of the indicated processing tasks performed by one or more processors.

[0105] It should be understood that according to some examples, a so-called "fluid" or flexible network concept may be adopted, wherein the operations and functions of a network element, a network function, or another entity of the network may be performed in a flexible manner in different entities or functions (such as nodes, hosts, or servers). In other words, the "division of labor" between the network elements, functions, or entities involved may vary from case to case.

[0106] Additionally, with respect to the bilateral model as described above, several challenges need to be addressed. In view of this, the underlying issues are outlined in more detail below.

[0107] That is, for bilateral models (term definition: a pair of (multiple) AI / ML models on which joint reasoning is performed, where joint reasoning includes AI / ML reasoning, whose reasoning is jointly performed across UE and network, i.e., the first part of the reasoning is first performed by UE, and then the remaining part is performed by gNB, and vice versa), since joint reasoning is required at both UE and network side parts, careful considerations on model training and model updating are required to ensure that the reasoning performance remains at a good level. To understand how bilateral model training is performed in a more practical setting, now refer to Figure 1 , which sets the context for the problems to be solved according to this specification.

[0108] refer to Figure 1 , the initial setup (training of bilateral models) can be done via offline engineering (i.e. may not be fully standardization related), where the network and UE vendors develop / train the corresponding network-side and UE-side parts (e.g. encoder and decoder) of the bilateral ML model via joint / separate model training. The dataset used for model training can be stored in the operator server so that any updates can be accessed by all parties. The exact format used for the dataset should be known or coordinated by the network and UE side.

[0109] according to Figure 1 , the following steps are considered. Model training is done based on offline engineering between UE vendor (101; 102; 103) and network vendor (111; 112; 113). Therefore, joint or separate training can be applied depending on the UE vendor. Additionally and / or alternatively, there may be multiple models, each corresponding to a different configuration / scenario and / or dataset (to support generalization).

[0110] In addition, (multiple) training datasets (e.g., for a CSI compression use case, for a given scenario / configuration, the dataset may be {e(H), H}. Here, H may be the channel and e(H) may be the compressed output of the encoder (for channel compression of H) may be updated and accessible by both the UE and the network provider. In another variant, (multiple) training datasets may be updated by only one party (e.g., the UE provider) and may be accessible by the network provider. Thus, this may allow training of newly available UE nodes and network nodes. Additionally, a neutral party (operator (121)), UE provider, or network provider may support storing the dataset.

[0111] In addition, using Figure 1 Given the assumed setup, it should be understandable that model training and deployment (i.e., considering the trained model) can occur without any model transmission and other complex air interface involvement, thereby saving air interface resources, as it should be understood that model transmission may typically require several MB of data, and the UE may not always be in a position to receive such a large amount of data (e.g., due to coverage).

[0112] However, in general, it is difficult to assume that a deployed ML model can work without any further updates over time, and similar statements are valid for bilateral models. The bilateral model used by the UE and gNB may also need to be updated / fine-tuned with newly acquired data in order to pursue good performance under changing radio conditions.

[0113] Therefore, reference Figure 2 and Figure 3 , the following two cases can be expected for model updating for the bilateral model, where Figure 2 shows an example of ML model or dataset update at the UE side for a bilateral model, and Figure 3 Examples of ML model or dataset updates at the network side for a bilateral model are shown.

[0114] 1. The UE-side part of the bilateral model (e.g., the encoder part corresponding to CSI compression) may be updated based on certain structural / architectural changes to the ML model and / or changes due to availability / updated datasets. Figure 2 An example is shown in which the UE performs a model update using newly available data and updates the dataset to dataset X2 in the remote server. Here, since the network is using a model that was not trained using dataset X2, the bilateral model may cause incompatibility and performance issues (e.g., information encoded by the UE using the updated model cannot be received, ultimately causing connection failure).

[0115] 2. The network-side portion of the dual model (e.g., the decoder portion corresponding to CSI compression) can be updated based on certain structural / architectural changes to the ML model and / or changes due to availability / updated datasets. Figure 3 An example is shown where the network performs a model update using newly available data (the update to the server is referred to as dataset X2). Here, since the UE is using a model that was not trained using dataset X2, the bilateral model may cause incompatibility and performance issues (e.g., unable to receive information encoded by the UE using the updated model, ultimately causing connection failure).

[0116] In view of the above, this specification proposes a NR framework to solve the above problems, especially Figure 2 Therefore, this specification solves the following two key problems.

[0117] That is, first, how the UE updates the ML model (e.g., notifies the network) when the underlying dataset changes, and second, how the network and UE communicate to perform ML model updates without sacrificing performance and latency.

[0118] Thus, the present specification allows the following advantages to be achieved by enabling the UE to update the ML model when the underlying data set changes, and by enabling communication between the network and the UE, which allows the ML model update to be performed without sacrificing performance and latency.

[0119] In the following, examples of embodiments are outlined which allow solving at least some of the above-mentioned problems and / or allowing achieving at least some of the above-mentioned advantages.

[0120] According to at least some examples of embodiments, when a bilateral model is used for joint model reasoning for a given feature / sub-use case / use case at the UE and network sides, and if the model is updated or needs to be updated at the UE, the following process may be followed to ensure that the model reasoning is consistent across the nodes involved in the bilateral model reasoning.

[0121] That is, the UE is configured (eg via higher layers) with parameters that enable / allow UE triggered / initiated UL transmissions in case any changes / updates / fine tuning are needed at the UE, at least for the UE part of the bilateral model.

[0122] The parameters enabling / permitting UE-triggered UL transmissions may include resource(s) to be used in the UL transmission, the format to be used in the UL transmission, and / or information to be carried in the UL transmission. For example, the UE-triggered UL transmission may be a scheduling request with dedicated PUCCH resources associated with a model update indication. In another example, the UE may be configured to use a dedicated MAC-CE command in the UL transmission so that the UE may indicate some information about the model update, such as whether the model update indication is for a model update or for approval of a model update, a version number of the used data set to be identified by the network, etc. Regarding the information to be carried in the UL transmission, the information may be a minimum level of information, which may be carried in at least one of the following, where in the event that the network may require additional details / information, more details / information may be requested later / subsequently by the network (by using, for example, additional and / or scheduled UL transmissions, as outlined in more detail below):

[0123] Information about the datasets used for / related to the model update (any relevant version numbers of the datasets used to be identified by the network);

[0124] Information about the reasoning complexity / latency of the UE part of the update of the bilateral model;

[0125] Information related to UE partial model configuration or output changes (output dimensions, capabilities);

[0126] any required model update duration (inference may not be applied for duration X until the model is fully updated) and / or the possibility of applying an older model during any model update duration; or

[0127] Any other relevant parameters associated with the bilateral model.

[0128] A UE-triggered / initiated model update may indicate a request for a model update or an approval for a model update at the UE. A request for a model update may mean that the UE is still able to perform joint reasoning based on an earlier version of the model, but the UE prioritizes updating the model based on the newly available data set. A UE indicating a request may represent the first scenario. Approval for a model update may apply when the UE changes the model from an earlier version of the model to an updated version without any pre-approval from the network, e.g., only the model applied at the UE is updated and the UE indicates the update to the network before using it for reasoning. A UE indicating an approval may represent the second scenario, where the UE performs a model update without consulting, e.g., the gNB, but simply indicates the update after the update is completed.

[0129] In addition, the UE determines that the model used at the UE may need to be updated (e.g., based on newly available data). The relevant data set used for model update / fine-tuning may be synchronized to a remote server (which is accessible to the network). In a secondary scenario (when approved, applicable), the UE may update the UE portion of the bilateral model with the newly available data.

[0130] In addition, the UE triggers / initiates UL transmission based on the received configuration according to the parameters enabling / allowing UL transmission and indicates a request for a model update. In a secondary scenario (when approval is applicable), the UE may also indicate that the UE portion of the bilateral model is updated.

[0131] In addition, based on the received UL transmission indicating a request (or approval) for a model update, the network may schedule an additional UL transmission (e.g., a PUSCH scheduled by a UL DCI), where the UL transmission may carry additional information related to the requested model update. The additional information may include:

[0132] Information related to the dataset used for / associated with the model update (any relevant version numbers of the dataset used to be identified by the network)

[0133] Information related to the reasoning complexity / latency of the UE part of the update of the bilateral model

[0134] Information related to UE partial model configuration or output changes (output dimensions, capabilities)

[0135] Any required model update duration (inference may not be applied for duration X until the model is fully updated) and / or the possibility of applying an older model during any model update duration

[0136] Any other relevant parameters associated with the bilateral model

[0137] Based on the scheduled UL transmission, the UE may send an UL transmission that provides more information about the request (or approval) for the model update.

[0138] Based on the received additional information related to the requested model update, the gNB may consider one or more of the following actions:

[0139] The requirement to obtain datasets from remote servers and view model updates on the network side,

[0140] Indicates whether the UE model can be updated (when a request for a model update is sent),

[0141] Deny approval for a model update (when approval for a model update is sent),

[0142] · suspending the use of the bilateral model during the model update duration (when a request for a model update is sent),

[0143] Instructs the UE to continue using the older version of the UE part of the bilateral model

[0144] Configure the necessary parameters to support joint reasoning with the updated UE part of the bilateral model

[0145] Trigger model updates at other nodes (other UEs) and network nodes based on newly available data

[0146] Reference now Figure 4 , Figure 4 Message sequences for bilateral ML model update scenarios according to various examples of embodiments are shown. Figure 4 The sequence will be outlined in detail.

[0147] Steps 1 to 5: It can be assumed that the UE 410 has an ML model trained using the dataset X1, and it has sent the ML model related capabilities to the network 420 in steps 2 and 3. The ML model capabilities in step 3 may indicate, for example, that the ML model is trained in the UE based on the dataset X1 (X1 may be a global or vendor-specific identifier identifying a batch of training / validation data), but the ML model may also be updated. Since it is a bilateral model, the network 420 may also obtain the dataset X1 from the operator's server as described above to train such a model. Since the ML model is updateable, the network prepares the ML model update configuration in the subsequent step 6.

[0148] Step 6: The network 420 configures the UE 410 to operate with the ML configuration corresponding to this trained model using the dataset X1. As part of the ML model update configuration, in one possible implementation, the network 420 configures RRC ASN.1 information elements (i.e., configuration parameters) to enable the UE 410 to trigger an UL transmission (e.g., the first UL transmission) in the event of any model update / change. The parameters enabling / allowing UE-triggered UL transmission may include the resource(s) to be used in the UL transmission, the format to be used in the UL transmission, and the information to be carried in the UL transmission. For example, the UE-triggered UL transmission may be a scheduling request with dedicated PUCCH resources associated with a model update indication.

[0149] Step 7: The network 420 and the UE 410 are aligned to operate on the ML model based on the dataset X1.

[0150] Step 8: At a later point in time, the ML model deployment function triggers the ML model update requirement.

[0151] UE 410 determines that the model used at UE 410 may need to be updated (e.g., based on newly available data)

[0152] • The relevant datasets used for model updating / fine-tuning can be synchronized to a remote server (which is accessible to the network).

[0153] • In the secondary variant (when approved for use), the UE 410 may update the UE portion of the bilateral model with the newly available data.

[0154] Steps 9, 10: This causes an ML model update request to the RRC layer in the UE 410, further triggering the message in step 10. The UE-triggered / initiated model update may indicate a request for a model update at the UE 410 or an approval for a model update. The request for a model update may mean that the UE 410 is still able to perform joint reasoning based on an earlier version of the model, but the UE 410 prioritizes updating the model based on the newly available dataset. In this case, this is dataset ID X2. This may represent the first scenario described above. When the UE 410 changes the model from an earlier version of the model to an updated version without any pre-approval from the network 420, the approval for the model update may apply, for example, only updating the model applied at the UE 410, and the UE 410 indicates the update to the network 420 before using it for reasoning. This may represent the second scenario described above.

[0155] Step 11: Based on the received UL transmission indicating a request (or approval) for a model update, the network 420 may schedule an additional UL transmission (e.g., a PUSCH scheduled by a UL DCI), wherein the UL transmission (e.g., a second UL transmission) may carry additional information related to the requested model update. The additional information may include:

[0156] Information related to the dataset used for / related to the model update (any relevant version number of the dataset used to be identified by the network)

[0157] Information related to the reasoning complexity / latency of the UE part of the update of the bilateral model

[0158] Information related to changes in some UE model outputs (dimensions, capabilities)

[0159] any required model update duration (inference may not be applied for duration X until the model is fully updated) and / or the possibility of applying an older model during any model update duration, any other relevant parameters associated with the bilateral model

[0160] Steps 12, 13: Based on the scheduled UL transmission, UE 410 may send an UL transmission providing more information about the request (or approval) for the model update. In a specific implementation, UE 410 sends the dataset ID X2 here as the updated dataset, which should be used for ML model training.

[0161] Steps 14 to 17: Based on the received additional information related to the requested model update. The gNB may indicate in the RRC configuration how the UE 410 should perform the ML update in the bilateral case (the information provided to the UE 410 based on, for example, the gNB performing one of the procedures outlined below may be understood as representing update information, where the UE may use such update information to understand how to proceed further, for example in view of updating (or not updating) the UE part of the bilateral model):

[0162] The requirement to obtain datasets from remote servers and view model updates on the network side,

[0163] Indicates to the UE 410 whether the model can be updated (when a request for a model update is sent),

[0164] Deny approval for a model update (when approval for a model update is sent),

[0165] · suspending the use of the bilateral model during the model update duration (when a request for a model update is sent),

[0166] Instruct UE 410 to continue using the older version of the UE part of the bilateral model

[0167] Trigger model updates at other nodes (other UEs) and network nodes based on newly available data

[0168] Also based on the information provided in step 11, the network 420 provides an RRC configuration, which includes updating the ML model configuration (e.g., updated dimensions). In addition, the network 420 also configures how the UE 410 should switch during the model update process. For example, the switch can be defined by a timer or an execution condition (i.e., switch immediately after the ML model update is completed or generate X output samples and then switch)

[0169] Steps 18 to 19: UE 410 may indicate to network 420 via the control plane (RRC) or user plane (e.g., in the payload of an uplink transmission, such as a CSI report) that it has successfully switched to the updated ML model in step 18. In step 19, both network 420 and UE 410 have switched to the updated ML model.

[0170] In the following, further examples of embodiments will be described in conjunction with the above-mentioned methods and / or apparatuses.

[0171] Reference now Figure 5 , shows a flowchart showing steps corresponding to the method according to various examples of embodiments. Such a method may include as described above with reference to Figure 4 At least some of the processes and / or steps outlined.

[0172] Specifically, according to Figure 5 , in S510, the method includes: obtaining a configuration with parameters at a terminal assigned to a network, the parameters enabling the terminal to initiate an uplink UL transmission, the uplink UL transmission being associated with an update at the terminal of at least a terminal portion of a bilateral model used at the terminal, wherein the bilateral model is used for joint model reasoning at the terminal side and the network side.

[0173] It should be noted that the method can be applied to the access network entity or function of the network to which the terminal is assigned. Such a network can be represented as reference Figure 4 In addition, such an access network entity or function may represent, for example, a gNB, as shown in FIG. Figure 4 The terminal communicating with the network may be represented by a terminal communicating with the gNB. In addition, such a terminal mentioned herein may represent a terminal as described in reference Figure 4 Such UE 410. In addition, Figure 5 Such step S510 may correspond to the above reference Figure 4 In addition, the bilateral model described in S510 may correspond to the one described in the above reference Figure 4 Such a bilateral model is described.

[0174] Furthermore, in S520 , the method includes: determining that an update of the bilateral model is required. Figure 5 Step S510 may correspond to the above reference Figure 4 At least a portion of step 8 as described above.

[0175] In addition, in S530, the method includes: initiating a first UL transmission based on the obtained configuration. Figure 5 Such step S530 may correspond to the above reference Figure 4 At least part of steps 9 and 10 as described above.

[0176] It should be noted that “initiate” may also be understood to mean “trigger”.

[0177] Furthermore, in S540, the method includes: sending an initiated first UL transmission, wherein the initiated first UL transmission indicates the required update. Figure 5Such step S540 may correspond to the above reference Figure 4 At least part of steps 9 and 10 as described above.

[0178] In addition, in S550, the method includes: receiving a response, the response including: update information related to the required update. Figure 5 Such step S550 may correspond to the above reference Figure 4 At least part of such steps 14 to 17 as described.

[0179] Furthermore, in S560, the method includes: establishing an update of the terminal part of the bilateral model based on the received response.

[0180] It should be noted that establishing an update may include updating the bilateral model and using the updated bilateral model for which approval has been obtained. Therefore, the term "establishing" includes both scenarios described above, the first scenario being that an update is performed based on a response, and the second scenario being that approval for the updated bilateral model is obtained. Figure 5 Such step S560 may correspond to the above reference Figure 4 At least part of such steps 14 to 17 as described.

[0181] In addition, according to at least some examples of embodiments, the first UL transmission may include a request for a bilateral model update. Such an UL transmission may correspond to the first scenario outlined above.

[0182] In addition, according to various examples of embodiments, the method may also include: updating the terminal part of the bilateral model based on the results obtained from the determination, wherein the first UL transmission includes: approval of the updated terminal part of the bilateral model, and wherein establishing includes: establishing the approved updated terminal part.

[0183] In addition, according to various examples of embodiments, the updates directly mentioned above may include: obtaining a second data set that can be used by the bilateral model from a remote server, the second data set being newer than the first data set currently used by the bilateral model; and updating the terminal part of the bilateral model based on the obtained second data set.

[0184] Optionally, according to at least some examples of the embodiments, the determining may further include: determining that a second data set that can be used by the bilateral model is provided at a remote server, the second data set being more recent than a first data set currently used by the bilateral model.

[0185] The (older / newer) dataset may also be understood as representing an (older / newer) version of (the terminal part and / or the network part of) the bilateral model.

[0186] In addition, according to various examples of embodiments, the parameter may include at least one of the following: radio resources to be used in the first UL transmission, data format and / or transmission format to be used in the first UL transmission, or information to be carried in the first UL transmission. The information includes at least one of the following: information related to a data set that can be used and / or is used to update the bilateral model, information related to reasoning complexity, the reasoning complexity is related to the terminal part of the bilateral model and / or the updated terminal part, information related to reasoning delay, the reasoning delay is related to the terminal part of the bilateral model and / or the updated terminal part, information related to the configuration of the terminal part of the bilateral model and / or the updated terminal part, information related to output changes of the bilateral model, the output changes of the bilateral model are associated with the terminal part of the bilateral model and / or the updated terminal part, the duration required to update the terminal part of the bilateral model, or the possibility of applying the terminal part of the bilateral model and / or the updated terminal part during the update duration for updating the bilateral model.

[0187] In addition, according to at least some examples of the embodiments, the method may also include: receiving a schedule for a second UL transmission in response to the sent first UL transmission; and sending the scheduled second UL transmission, the scheduled second UL transmission including at least one of the following information related to the required update: information related to a data set that can be used and / or is used to update the bilateral model, information related to reasoning complexity, reasoning complexity is related to the terminal part and / or the updated terminal part of the bilateral model, information related to reasoning latency, reasoning latency is related to the terminal part and / or the updated terminal part of the bilateral model, information related to the configuration of the terminal part and / or the updated terminal part of the bilateral model, information related to output changes of the bilateral model, output changes of the bilateral model are associated with the terminal part and / or the updated terminal part of the bilateral model, the duration required to update the terminal part of the bilateral model, or the possibility of applying the terminal part and / or the updated terminal part of the bilateral model during the update duration for updating the bilateral model. This may correspond to the above reference Figure 4 At least part of such steps 11 to 13 are outlined.

[0188] In addition, according to various examples of the embodiments, the method may further include: in response to the sent first UL transmission or the second UL transmission, receiving a response including update information, wherein the update information includes at least one of the following: if the first UL transmission includes a request, an indication of whether the terminal part of the bilateral model can be updated, if the first UL transmission includes a request, an indication of whether to suspend the use of the bilateral model during an update duration for updating the bilateral model, if the first UL transmission includes a request, a configuration for updating the terminal part of the bilateral model, if the first UL transmission includes an approval, an indication of whether the approval is rejected, a configuration of how to switch within the update duration for updating the bilateral model, or an indication of continuing to use the bilateral model without updating the bilateral model.

[0189] In addition, according to various example embodiments, the method may further include: switching to an updated bilateral model based on the received response; and indicating the switching to the updated bilateral model via the control plane or the user plane. This may correspond to the above reference Figure 4 At least part of such steps 18 and 19 are outlined.

[0190] Optionally, according to at least some examples of embodiments, the bilateral model can be a bilateral AI / ML model.

[0191] The above solution allows terminal (respectively UE) initiated model update for bilateral AI / MI model. Therefore, the advantage of the above solution is that it enables efficient and / or secure and / or robust and / or failure-resistant and / or flexible terminal (respectively UE) initiated model update for bilateral AI / MI model.

[0192] Reference now Figure 6 , Figure 6 A flowchart showing steps corresponding to the method according to various examples of embodiments is shown. The method may include at least some of the above reference Figure 4 Outline the process and / or steps.

[0193] Specifically, according to Figure 6 , in S610, the method includes: receiving a first uplink UL transmission from a terminal assigned to the network at an access network entity or function of the network, the first uplink UL transmission indicating a need for an update of at least a terminal part of a bilateral model used at the terminal, wherein the bilateral model is used for joint model reasoning on the terminal side and the network side.

[0194] Figure 6 Such step S610 may correspond to the above reference Figure 4 At least part of such steps 9 and 10 are outlined.

[0195] In addition, according to Figure 6 In 620, the method includes sending a response, the response including update information related to the required update.

[0196] Figure 6 Such step S620 may correspond to the above reference Figure 4 At least part of such steps 14 to 17 are outlined.

[0197] In addition, according to various examples of embodiments, the method may further include: providing a configuration with parameters to the terminal, the parameters enabling the terminal to initiate a first UL transmission, wherein the parameters include at least one of the following: radio resources to be used in the first UL transmission, data formats and / or transmission formats to be used in the first UL transmission, or information to be carried in the first UL transmission. The information includes at least one of the following: information related to a data set that can be used and / or used to update the bilateral model, information related to reasoning complexity, reasoning complexity related to the terminal part and / or the updated terminal part of the bilateral model, information related to reasoning delay, reasoning delay related to the terminal part and / or the updated terminal part of the bilateral model, information related to the configuration of the terminal part and / or the updated terminal part of the bilateral model, information related to output changes of the bilateral model, output changes of the bilateral model are associated with the terminal part and / or the updated terminal part of the bilateral model, the duration required to update the terminal part of the bilateral model, or the possibility of applying the terminal part and / or the updated terminal part of the bilateral model during the update duration for updating the bilateral model.

[0198] In addition, according to at least some examples of the embodiments, the method may also include: scheduling a second UL transmission in response to the received first UL transmission; and receiving the scheduled second UL transmission, the scheduled second UL transmission including at least one of the following information related to the required update: information related to a data set that can be used and / or used to update the bilateral model, information related to the reasoning complexity related to the terminal part and / or the updated terminal part of the bilateral model, information related to the reasoning delay related to the terminal part and / or the updated terminal part of the bilateral model, information related to the configuration of the terminal part and / or the updated terminal part of the bilateral model, information related to the output change of the bilateral model associated with the terminal part and / or the updated terminal part of the bilateral model, the duration required to update the terminal part of the bilateral model, or the possibility of applying the terminal part and / or the updated terminal part of the bilateral model during the update duration for updating the bilateral model.

[0199] In addition, according to various examples of the embodiment, the method may also include: in response to the received first or second UL transmission, performing at least one of the following: obtaining a data set for updating the bilateral model from a remote server, and obtaining a network side requirement for updating the bilateral model, updating the network side of the bilateral model, if the first UL transmission includes a request for updating the bilateral model, indicating to the terminal whether the terminal part of the bilateral model is capable of being updated, if the first UL transmission includes approval for the terminal part of the update of the bilateral model, determining whether to reject the approval, if the first UL transmission includes a request for updating the bilateral model, suspending the use of the bilateral model during the update duration for updating the bilateral model, indicating to the terminal to continue using the bilateral model without updating the bilateral model, configuring parameters related to the bilateral model to support joint reasoning with the updated terminal part of the bilateral model, or triggering an update of the bilateral model at another access network entity or function and / or another terminal.

[0200] In addition, according to various examples of the embodiment, the method may further include: receiving an indication that the terminal has switched to an updated bilateral model through a control plane or a user plane.

[0201] The above solution allows terminal (UE) initiated model update for bilateral AI / MI model. Therefore, the advantage of the above solution is that it enables efficient and / or secure and / or robust and / or fault-resistant and / or flexible terminal (respectively UE) initiated model update for bilateral AI / MI model.

[0202] Reference now Figure 7 , Figure 7 A block diagram illustrating an apparatus according to various examples of embodiments is shown.

[0203] Specifically, Figure 7 A block diagram of an apparatus 700 according to various examples of embodiments is shown, which apparatus 700 may represent a terminal or UE, for example, as described above with reference to Figure 4 Such a UE 410 is outlined, which can participate in the model update for the bilateral AI / MI model initiated by the terminal / UE. In addition, even if the terminal is mentioned, the terminal can also be another device or function with similar tasks, such as a chipset, a chip, a module, an application, etc., which can also be part of a network element or attached to a network element as a separate element, etc. It should be understood that each block and any combination thereof can be implemented by various components or combinations thereof, such as hardware, software, firmware, one or more processors and / or circuit systems.

[0204] Figure 7The device 700 shown may include a processing circuit system, a processing function, a control unit or a processor 710, such as a CPU, etc., which is suitable for enabling a model update for a bilateral AI / MI model initiated by a terminal / UE. The processor 710 may include one or more processing parts or functions dedicated to a specific process as described below, or the process may be executed in a single processor or processing function. For example, a portion for performing such a specific process may also be provided as a discrete element, or provided in one or more additional processors, processing functions or processing parts, such as in a physical processor (such as a CPU) or in one or more physical or virtual entities. Reference numerals 731 and 732 represent input / output (I / O) units or functions (interfaces) connected to the processor or processing function 710. The I / O units 731 and 732 may be a combined unit including a communication device for several entities / elements, or may include a distributed structure with multiple different interfaces for different entities / elements. Reference numeral 720 represents a memory, which may be used, for example, to store data and programs to be executed by the processor or processing function 710 and / or as a working storage for the processor or processing function 710. It should be noted that the memory 720 may be implemented by using one or more memory portions of the same or different types of memory, but may also represent an external memory, such as an external database provided on a cloud server.

[0205] The processor or processing function 710 is configured to perform processing related to the above-mentioned processing. Specifically, the processor or processing circuit system or function 710 includes one or more of the following sub-parts. Sub-part 711 is an acquisition part, which can be used as a part for obtaining configuration. Part 711 can be configured according to Figure 5 S510 performs processing. In addition, sub-section 712 is a determination section, which can be used as a section for determining the update requirement. Section 712 can be configured according to Figure 5 S520 performs processing. In addition, sub-section 713 is an initiating section, which can be used as a section for initiating UL transmission. Section 713 can be configured according to Figure 5 S530 performs processing. Subsection 714 is a transmission section, which can be used as a section for transmitting an initiated UL transmission. Section 714 can be configured according to Figure 5 S540 performs processing. In addition, sub-section 715 is a receiving section, which can be used as a section for receiving a response. Section 715 can be configured according to Figure 5 S550 performs processing. In addition, sub-section 716 is a setup section, which can be used as a section for establishing updates. Section 716 can be configured according to Figure 5 The processing of S560 is executed.

[0206] Reference now Figure 8 , Figure 8 A block diagram illustrating an apparatus according to various examples of embodiments is shown.

[0207] Specifically, Figure 8 A block diagram showing an apparatus according to various examples of embodiments is shown, which apparatus may represent an access network entity or function, such as described above with reference to Figure 4 Such a gNB is outlined, which can participate in the model update initiated by the terminal / UE for the bilateral AI / MI model. In addition, even if the access network entity or function is mentioned, the access network entity or function can also be another device or function with similar tasks, such as a chipset, chip, module, application, etc., which can also be part of the network element or attached to the network element as a separate element, etc. It should be understood that each block and any combination thereof can be implemented by various components or combinations thereof, such as hardware, software, firmware, one or more processors and / or circuit systems.

[0208] Figure 8 The device 800 shown in the figure may include a processing circuit system, a processing function, a control unit or a processor 810, such as a CPU, etc., which is suitable for enabling a model update for a bilateral AI / MI model initiated by a terminal / UE. The processor 810 may include one or more processing parts or functions dedicated to a specific process, as described below, or the process may be executed in a single processor or processing function. For example, the part for performing such a specific process may also be provided as a discrete element, or provided in one or more additional processors, processing functions or processing parts, such as in a physical processor (such as a CPU), or in one or more physical or virtual entities. Reference numerals 831 and 832 represent input / output (I / O) units or functions (interfaces) connected to the processor or processing function 810. The I / O units 831 and 832 may be a unit including a combination of communication devices for several entities / elements, or may include a distributed structure with multiple different interfaces for different entities / elements. Reference numeral 820 represents a memory, which can be used, for example, to store data and programs to be executed by the processor or processing function 810 and / or as a working storage for the processor or processing function 810. It should be noted that the memory 820 may be implemented by using one or more memory portions of the same or different types of memory, but may also represent external memory, such as an external database provided on a cloud server.

[0209] The processor or processing function 810 is configured to perform processing related to the above-mentioned processing. Specifically, the processor or processing circuit system or function 810 includes one or more of the following sub-parts. Sub-part 811 is a receiving part, which can be used as a part for receiving UL transmissions. Part 811 can be configured according to Figure 6 S610 performs processing. In addition, sub-section 812 is a sending section, which can be used as a section for sending a response. Section 812 can be configured according to Figure 6 The process of S620 is executed.

[0210] It should be noted that, as mentioned above, Figure 7 and 8 The apparatuses 700 and 800 described above may include additional / additional sub-parts that may allow the apparatuses 700 and 800 to perform the above-referenced Figure 4 Such method / method steps as described.

[0211] It should be understood

[0212] -The access technology via which traffic is transmitted to and from entities in the communication network can be any suitable current or future technology, such as WLAN (Wireless Local Access Network), WiMAX (Worldwide Interoperability for Microwave Access), LTE, LTE-A, 5G, Bluetooth, infrared, etc. can be used; in addition, embodiments can also apply wired technology, such as IP-based access technology, such as a wired network or a fixed line.

[0213] -Embodiments suitable for implementation as software code or portion thereof and running using a processor or processing functionality are independent of software code and may be specified using any known or future developed programming language, such as a high-level programming language such as objective-C, C, C++, C#, Java, Python, Javascript, other scripting languages, etc., or a low-level programming language such as machine language or assembly language.

[0214] -The implementation of the embodiments is hardware independent and can be implemented using any known or future developed hardware technology or any mixture of these technologies, such as a microprocessor or CPU (Central Processing Unit), MOS (Metal Oxide Semiconductor), CMOS (Complementary MOS), BiMOS (Bipolar MOS), BiCMOS (Bipolar CMOS), ECL (Emitter Coupled Logic) and / or TTL (Transistor-Transistor Logic).

[0215] - embodiments may be implemented as separate devices, means, units, components or functions, or in a distributed manner, for example, one or more processors or processing functions may be used or shared in a process, or one or more processing parts or processing parts may be used and shared in a process, where one physical processor or more than one physical processor may be used to implement one or more processing parts dedicated to the specific process,

[0216] - the device may be implemented by a semiconductor chip, a chipset or a (hardware) module including such a chip or chipset;

[0217] - The embodiments may also be implemented as any combination of hardware and software, such as ASIC (Application Specific IC (Integrated Circuit)) components, FPGA (Field Programmable Gate Array) or CPLD (Complex Programmable Logic Device) components or DSP (Digital Signal Processor) components.

[0218] The embodiments may also be implemented as a computer program product, including a computer usable medium having a computer readable program code embodied therein, the computer readable program code being adapted to execute the processes as described in the embodiments, wherein the computer usable medium may be a non-transitory medium.

[0219] Although the present disclosure has been described herein before with reference to specific embodiments thereof, the present disclosure is not limited thereto and various modifications may be made thereto.

Claims

1. A method comprising: obtaining (S510) at a terminal assigned to a network a configuration with parameters enabling the terminal to initiate an uplink (UL) transmission, the uplink (UL) transmission being associated with an update at the terminal for at least a terminal part of a bilateral model used at the terminal, wherein the bilateral model is used for joint model reasoning at the terminal side and at the network side; determining ( S520 ) that an update of the bilateral model is required; Based on the obtained configuration, initiating (S530) a first UL transmission; sending (S540) the first UL transmission initiated, the first UL transmission initiated indicating the required update; receiving (S550) a response, the response comprising: update information related to the required update; as well as Based on the received response, an update of the terminal portion of the bilateral model is established (S560).

2. The method according to claim 1, The first UL transmission includes: A request for an update of the bilateral model.

3. The method according to claim 1 or 2, further comprising: updating the terminal portion of the bilateral model based on a result obtained from the determining, wherein the first UL transmission includes: approval of the terminal portion of the update of the bilateral model, and The establishing includes: establishing an approved terminal portion of the update.

4. The method according to claim 3, The updates include: Obtaining from a remote server a second data set that can be used by the bilateral model, the second data set being more recent than a first data set currently used by the bilateral model; as well as Based on the obtained second data set, the terminal part of the bilateral model is updated.

5. The method according to any one of claims 1 to 4, The determination includes: It is determined that a second data set usable by the bilateral model is provided at a remote server, the second data set being more recent than a first data set currently used by the bilateral model.

6. The method according to any one of claims 1 to 5, The parameters include at least one of the following: a radio resource to be used in the first UL transmission, a data format and / or a transmission format to be used in the first UL transmission, or The information to be carried in the first UL transmission, wherein the information includes at least one of the following: information related to a data set that can be used and / or is used to update the bilateral model, information related to reasoning complexity associated with the terminal part and / or an updated terminal part of the bilateral model, information related to an inference delay associated with the terminal part and / or an updated terminal part of the bilateral model, information relating to the configuration of the terminal part and / or an updated terminal part of the bilateral model, information related to a change in output of the bilateral model, the change in output of the bilateral model being associated with the terminal portion and / or an updated terminal portion of the bilateral model, The duration required to update the terminal part of the bilateral model, or the possibility to apply the terminal part of the bilateral model and / or an updated terminal part during an update duration for updating the bilateral model.

7. The method according to any one of claims 1 to 6, further comprising: receiving a schedule for a second UL transmission in response to the transmitted first UL transmission; as well as sending the second scheduled UL transmission, the second scheduled UL transmission including at least one of the following information related to the required update: information related to a data set that can be used and / or is used to update the bilateral model, information related to reasoning complexity associated with the terminal part and / or an updated terminal part of the bilateral model, information related to an inference delay associated with the terminal part and / or an updated terminal part of the bilateral model, information relating to the configuration of the terminal part and / or an updated terminal part of the bilateral model, information related to a change in output of the bilateral model, the change in output of the bilateral model being associated with the terminal portion and / or an updated terminal portion of the bilateral model, The duration required to update the terminal part of the bilateral model, or the possibility to apply the terminal part of the bilateral model and / or an updated terminal part during an update duration for updating the bilateral model.

8. The method according to any one of claims 1 to 7, further comprising: In response to the first UL transmission or the second UL transmission being sent, receiving the response including the update information, wherein the update information includes at least one of the following: an indication of whether the terminal part of the bilateral model can be updated if the first UL transmission includes the request, if the first UL transmission includes the request, an indication of whether to suspend use of the bilateral model during an update duration for updating the bilateral model, if the first UL transmission comprises the request, comprising an updated configuration of the terminal part of the bilateral model, if the first UL transmission includes the grant, an indication of whether the grant is rejected, a configuration on how to switch within the update duration for updating the bilateral model, or An indication of continuing to use the bilateral model without updating the bilateral model.

9. The method according to any one of claims 1 to 8, further comprising: Based on the received response, switching to an updated bilateral model; as well as The switching to the updated bilateral model is indicated via a control plane or a user plane.

10. The method according to any one of claims 1 to 9, The bilateral model is a bilateral artificial intelligence / machine learning AI / ML model.

11. A method comprising: receiving (S610), at an access network entity or function of a network, a first uplink UL transmission from a terminal assigned to the network, the first uplink UL transmission indicating a need for an update of at least a terminal part of a bilateral model used at the terminal, wherein the bilateral model is used for joint model reasoning at the terminal side and the network side; as well as A response is sent (S620), the response comprising: update information related to the required update.

12. The method according to claim 11, further comprising: providing a configuration with parameters to the terminal, the parameters enabling the terminal to initiate the first UL transmission, The parameters include at least one of the following: a radio resource to be used in the first UL transmission, a data format and / or a transmission format to be used in the first UL transmission, or The information to be carried in the first UL transmission, wherein the information includes at least one of the following: information related to a data set that can be used and / or is used to update the bilateral model, information related to reasoning complexity associated with the terminal part and / or an updated terminal part of the bilateral model, information related to an inference delay associated with the terminal part and / or an updated terminal part of the bilateral model, information relating to the configuration of the terminal part and / or an updated terminal part of the bilateral model, information related to a change in output of the bilateral model, the change in output of the bilateral model being associated with the terminal portion and / or an updated terminal portion of the bilateral model, the duration required to update the terminal portion of the bilateral model, or A possibility to apply the terminal part and / or an updated terminal part of the bilateral model during an update duration for updating the bilateral model.

13. The method according to claim 11 or 12, further comprising: In response to the received first UL transmission, scheduling a second UL transmission; as well as receiving the second scheduled UL transmission, the second scheduled UL transmission including at least one of the following information related to the required update: information related to a data set that can be used and / or is used to update the bilateral model, information related to reasoning complexity associated with the terminal part and / or an updated terminal part of the bilateral model, information related to an inference delay associated with the terminal part and / or an updated terminal part of the bilateral model, information relating to the configuration of the terminal part and / or an updated terminal part of the bilateral model, information related to a change in output of the bilateral model, the change in output of the bilateral model being associated with the terminal portion and / or an updated terminal portion of the bilateral model, the duration required to update the terminal portion of the bilateral model, or A possibility to apply the terminal part and / or an updated terminal part of the bilateral model during an update duration for updating the bilateral model.

14. The method according to any one of claims 11 to 13, further comprising: In response to the received first UL transmission or the received second UL transmission, perform at least one of the following: Obtaining a data set for updating the bilateral model from a remote server, and obtaining a requirement on the network side for updating the bilateral model, updating the network side of the bilateral model, If the first UL transmission includes a request for an update of the bilateral model, indicating to the terminal whether the terminal part of the bilateral model is updateable, If the first UL transmission includes an approval for the terminal portion of the update of the bilateral model, determining whether to reject the approval, If the first UL transmission includes a request for an update of the bilateral model, suspending use of the bilateral model during an update duration for updating the bilateral model, indicating to the terminal to continue to use the bilateral model without updating the bilateral model, configuring parameters associated with the bilateral model to support joint reasoning with an updated terminal portion of the bilateral model, or An update of the bilateral model at another access network entity or function and / or another terminal is triggered.

15. The method according to any one of claims 11 to 14, further comprising: An indication that the terminal has switched to the updated bilateral model is received through a control plane or a user plane.

16. An apparatus (700), comprising: at least one processor; as well as at least one memory storing instructions which, when executed by the at least one processor, cause the apparatus (700) to at least: Obtaining a configuration having parameters that enable the device (700) assigned to the network to initiate an uplink (UL) transmission associated with an update at the device (700) of at least a device portion of a bilateral model used at the device (700), wherein the bilateral model is used for joint model reasoning at the device side and the network side; determining that an update of the bilateral model is required; Initiate a first UL transmission based on the obtained configuration; sending the first initiated UL transmission, the first initiated UL transmission indicating the required update; receiving a response, the response comprising: update information related to the desired update; as well as Based on the received response, an update of the device portion of the bilateral model is established.

17. The device (700) according to claim 16, The first UL transmission includes: A request for an update of the bilateral model.

18. The apparatus (700) according to claim 16 or 17, wherein the apparatus is further caused to: updating the device portion of the bilateral model based on a result obtained from the determining, The first UL transmission includes: approval of the device portion of the update for the bilateral model, and Wherein said establishing comprises: said device is also caused to establish an approved device portion of said update.

19. The device (700) according to claim 18, wherein the updating comprises the device further being caused to: Obtaining from a remote server a second data set that can be used by the bilateral model, the second data set being more updated than a first data set currently used by the bilateral model; and Based on the obtained second data set, the device portion of the bilateral model is updated.

20. The device (700) according to any one of claims 16 to 19, wherein the determining comprises the apparatus further causing: It is determined that a second data set usable by the bilateral model is provided at a remote server, the second data set being more recent than a first data set currently used by the bilateral model.

21. The device (700) according to any one of claims 16 to 20, The parameters include at least one of the following: a radio resource to be used in the first UL transmission, a data format and / or a transmission format to be used in the first UL transmission, or The information to be carried in the first UL transmission, wherein the information includes at least one of the following: information related to a data set that can be used and / or is used to update the bilateral model, information related to reasoning complexity associated with the device portion and / or updated device portion of the bilateral model, information related to an inference delay associated with the device portion and / or an updated device portion of the bilateral model, information relating to the configuration of the device part and / or updated device part of the bilateral model, information related to a change in an output of the bilateral model, the change in the output of the bilateral model being associated with the device portion and / or an updated device portion of the bilateral model, the duration required to update the device portion of the bilateral model, or A possibility to apply the device part and / or an updated device part of the bilateral model during an update duration for updating the bilateral model.

22. The apparatus (700) according to any one of claims 16 to 21, wherein the apparatus is further caused to: receiving a schedule for a second UL transmission in response to the sent first UL transmission; and sending the second scheduled UL transmission, the second scheduled UL transmission including at least one of the following information related to the required update: information related to a data set that can be used and / or is used to update the bilateral model, information related to reasoning complexity associated with the device portion and / or updated device portion of the bilateral model, information related to an inference delay associated with the device portion and / or an updated device portion of the bilateral model, information relating to the configuration of the device part and / or updated device part of the bilateral model, information related to a change in an output of the bilateral model, the change in the output of the bilateral model being associated with the device portion and / or an updated device portion of the bilateral model, the duration required to update the device portion of the bilateral model, or A possibility to apply the device part and / or an updated device part of the bilateral model during an update duration for updating the bilateral model.

23. The device (700) according to any one of claims 16 to 22, wherein the device is further configured to: In response to the first UL transmission or the second UL transmission being sent, receiving the response including the update information, wherein the update information includes at least one of the following: an indication of whether the device portion of the bilateral model is updateable if the first UL transmission includes the request, if the first UL transmission includes the request, an indication of whether to suspend use of the bilateral model during an update duration for updating the bilateral model, if the first UL transmission includes the request, then including an updated configuration of the device portion of the bilateral model, if the first UL transmission includes the grant, an indication of whether the grant is rejected, a configuration on how to switch within the update duration for updating the bilateral model, or An indication of continuing to use the bilateral model without updating the bilateral model.

24. The device (700) according to any one of claims 16 to 23, wherein the device is further configured to: Based on the received response, switching to an updated bilateral model; and The switching to the updated bilateral model is indicated via a control plane or a user plane.

25. The device (700) according to any one of claims 16 to 24, The bilateral model is a bilateral artificial intelligence / machine learning AI / ML model.

26. An apparatus (800), comprising: at least one processor; as well as at least one memory storing instructions which, when executed by the at least one processor, cause the apparatus (800) to at least: The device (800) is an access network entity or function of the network, receiving a first uplink (UL) transmission from a terminal assigned to the network, the first uplink (UL) transmission indicating a need for an update of at least a terminal part of a bilateral model used at the terminal, wherein the bilateral model is used for joint model reasoning at the terminal side and the network side; as well as A response is sent, the response comprising: update information related to the desired update.

27. The apparatus (800) of claim 26, wherein the apparatus is further caused to: providing a configuration with parameters to the terminal, the parameters enabling the terminal to initiate the first UL transmission, The parameters include at least one of the following: a radio resource to be used in the first UL transmission, a data format and / or a transmission format to be used in the first UL transmission, or The information to be carried in the first UL transmission, wherein the information includes at least one of the following: information related to a data set that can be used and / or is used to update the bilateral model, information related to reasoning complexity associated with the terminal part and / or an updated terminal part of the bilateral model, information related to an inference delay associated with the terminal part and / or an updated terminal part of the bilateral model, information relating to the configuration of the terminal part and / or an updated terminal part of the bilateral model, information related to a change in output of the bilateral model, the change in output of the bilateral model being associated with the terminal portion and / or an updated terminal portion of the bilateral model, the duration required to update the terminal portion of the bilateral model, or A possibility to apply the terminal part and / or an updated terminal part of the bilateral model during an update duration for updating the bilateral model.

28. The apparatus (800) according to claim 26 or 27, wherein the apparatus is further caused to: In response to the received first UL transmission, scheduling a second UL transmission; and receiving the second UL transmission scheduled, the second UL transmission scheduled including at least one of the following information related to the required update: information related to a data set that can be used and / or is used to update the bilateral model, information related to reasoning complexity associated with the terminal part and / or an updated terminal part of the bilateral model, information related to an inference delay associated with the terminal part and / or an updated terminal part of the bilateral model, information relating to the configuration of the terminal part and / or an updated terminal part of the bilateral model, information related to a change in output of the bilateral model, the change in output of the bilateral model being associated with the terminal portion and / or an updated terminal portion of the bilateral model, the duration required to update the terminal portion of the bilateral model, or A possibility to apply the terminal part and / or an updated terminal part of the bilateral model during an update duration for updating the bilateral model.

29. The apparatus (800) of any one of claims 26 to 28, wherein the apparatus is further caused to: in response to receiving the first UL transmission or the second UL transmission, perform at least one of the following: Obtaining a data set for updating the bilateral model from a remote server, and obtaining a requirement on the network side for updating the bilateral model, updating the network side of the bilateral model, If the first UL transmission includes a request for an update of the bilateral model, indicating to the terminal whether the terminal part of the bilateral model is updateable, If the first UL transmission includes an approval for the terminal portion of the update of the bilateral model, determining whether to reject the approval, If the first UL transmission includes a request for an update of the bilateral model, suspending use of the bilateral model during an update duration for updating the bilateral model, indicating to the terminal to continue to use the bilateral model without updating the bilateral model, configuring parameters associated with the bilateral model to support joint reasoning with an updated terminal portion of the bilateral model, or An update of the bilateral model at another access network entity or function and / or another terminal is triggered.

30. The device (800) according to any one of claims 26 to 29, wherein the device is further caused to: An indication that the terminal has switched to an updated bilateral model is received through a control plane or a user plane.

31. A computer program product for a computer, comprising software code portions for performing the steps of any one of claims 1 to 10 or any one of claims 11 to 15 when said product is run on said computer.

32. The computer program product of claim 31, wherein The computer program product comprises a computer readable medium on which the software code portions are stored, and / or The computer program product can be directly loaded into the internal memory of the computer and / or can be sent via a network via at least one of an upload, download and push process.

Images