Capability signaling for machine learning based coding of uplink control information

Abstract

A wireless device (10) receives control data provided by a node (100) of a wireless communication network. The control data indicates a set of one or more supported machine learning models for encoding and decoding uplink control signaling from the wireless device (10) to the node (100). The wireless device (10) selects one or more of the machine learning models from the set. Based on at least one of the one or more selected machine learning models, the wireless device (10) encodes uplink control signaling from the wireless device (10) to the node (100).

Description

Technical Field

[0001] The present disclosure relates to methods for managing network services and corresponding devices, systems, and computer programs. Background Art

[0002] In wireless communication networks, for example, as specified by 3GPP (3rd Generation Partnership Project), efficient coding of wireless transmissions is an important factor. For example, the 5th generation (5G) technology specified by 3GPP, also denoted as NR (New Radio) technology, uses OFDM (Orthogonal Frequency Division Multiplexing) with configurable bandwidth and subcarrier spacing to efficiently support a diverse set of use cases and deployment scenarios. With respect to the LTE (Long Term Evolution) technology specified by 3GPP, the NR technology provides improvements with respect to deployment flexibility, achievable user throughput, latency and reliability. The NR technology also provides enhanced support for spatial multiplexing, in particular spatial sharing of time-frequency resources across users, commonly referred to as Multi-User MIMO (MU-MIMO).

[0003] Figure 1 Schematic diagram showing the principle of MU-MIMO transmission: with N TX A multi-antenna base station with multiple antenna ports sends information to several UEs simultaneously (i.e., on the same OFDM time-frequency resources). () is transmitted to UE(1), sequence S () is transmitted to UE (2) etc. Before modulation and transmission, for each sequence S () Application Precoding to spatially separate transmissions and mitigate multiplexing interference.

[0004] Each UE demodulates its received signal and combines the received antenna signals to obtain an estimate of the transmitted sequence The estimate It can be expressed as

[0005]

[0006] The second addition term in (1) represents the spatial multiplexing interference seen by UE(i). The goal of the base station is to construct the precoder set In this way: norm is large, and the norm ≠i is small. In other words, the precoder The channel H observed by UE(i) should be provided () , while providing low correlation with the channels observed by other UEs. In order to construct a precoder for efficient MU-MIMO transmission, the base station usually needs to obtain detailed knowledge of the channel H(i).

[0007] In deployments where channel reciprocity holds, detailed channel knowledge can be obtained from uplink sounding reference signals (SRS) sent periodically or on demand by active UEs. Based on these SRS, the base station can directly estimate the uplink channel and, hence, the downlink channel H(i).

[0008] However, if channel reciprocity does not hold, the active UE needs to feed back channel details to the base station. This feedback information is also called channel state information (CSI). In LTE and NR, the feedback of CSI is typically based on a CSI reference signal (CSI-RS) periodically sent by the base station. From the CSI-RS, the UE can estimate the downlink channel and then report the corresponding CSI to the base station. CSI is typically reported on an uplink control channel or on an uplink data channel. The base station then uses the reported CSI to select a suitable precoder for downlink MU-MIMO transmission.

[0009] The CSI feedback mechanism for MU-MIMO operation in NR is called CSI Type II. Type IICSI is based on specifying a set of discrete Fourier transform (DFT) basis functions, sometimes also denoted as a "beam grid", from which the UE selects those that best match its channel conditions, similar to the classic codebook PMI (precoding matrix indicator). In addition, the UE also reports how these beams should be combined in terms of relative amplitude scaling and co-phasing. The number of beams reported by the UE is configurable and can typically be 2 or 4. Reporting more beams increases the CSI resolution feedback, but at the expense of additional uplink signaling overhead.

[0010] Figure 2 Further explanation of CSI Type II. As shown in the figure, DFT beam vector b is realized from a wideband perspective. n The choice of and corresponding to the selected DFT beam vector b n The relative amplitude a n The phase angle θ of each sub-band is determined by n Here, “wideband perspective” means that the selected DFT beam vector is the same for all subcarriers used in OFDM transmission, while “per subband” means that the common phase parameters are determined over a subset of consecutive subcarriers. The common phase parameters are quantized such that Taken from the QPSK or 8PSK signal constellation. Where k represents the subband index, the precoder reported by the UE can be expressed as

[0011]

[0012] When a base station spatially schedules multiple UEs, it can select UEs that have reported different sets of beams or beams with weak correlation. CSI Type II thus represents a MIMO channel feedback mechanism where the UE reports a precoder assumption that trades off CSI resolution with uplink signaling overhead. Given this situation, the use of CSI Type II can be viewed as a suboptimal precoding method for MU-MIMO that attempts to balance uplink signaling overhead with MU-MIMO precoding performance.

[0013] Recently, neural network-based autoencoders (AEs) have shown promising results for downlink MIMO channel estimation for compressed uplink feedback. AEs are a type of artificial neural network (NN) that can be used to compress data in an unsupervised manner. That is, AEs work only with input data and do not require output labels. These networks aim to reconstruct the input data at the output layer with high fidelity relative to a given loss function.

[0014] Figure 3 A simple fully connected or dense AE architecture is schematically shown. With such an AE architecture, the AE can be considered to be divided into two parts: an encoder located at the UE, and a decoder located at the base station. The output of the encoder corresponding to the bottleneck layer of the NN represents the code value to be signaled from the UE to the base station.

[0015] AE can have different architectures. For example, AE can be based on fully connected NN, multi-dimensional convolutional NN, recursive NN or any combination thereof. However, all AE architectures have Figure 3 The encoder-bottleneck-decoder structure shown.

[0016] When using AE, Figure 3 The size of the code to be signaled, represented by Y in Figure 3 The size of the input data represented by X in . The encoder part of the AE reduces the spatial dimensionality of the input features as the depth of the NN increases. The decoder part of the AE basically does the opposite, i.e., it gradually returns the compressed codes to their original feature size. At the output layer, the AE reconstructs the original input data with some loss. Figure 3 In the above example, the reconstructed data is given by express.

[0017] The architecture of an AE is often numerically optimized for a specific application, such as channel compression. This is done through a process called hyperparameter tuning. When optimizing the AE architecture, the hardware limitations of the encoder and decoder also need to be considered.

[0018] The weights and biases of the AE (with a fixed architecture) can be optimized to improve the reconstructed data For example, the weights and biases can be trained to minimize the mean squared error (MSE) of the reconstruction, i.e., to minimize Model training is typically done using some variation of stochastic gradient descent on a large dataset. This dataset should represent the actual data that the AE will encounter during field operations.

[0019] The process of designing an AE, especially hyperparameter tuning and model training, is typically expensive and consumes a lot of computing, storage, and power resources. In addition, in order to achieve good performance, the encoder and decoder parts of the AE may need to be trained jointly. This may cause scalability issues because the decoder part at the base station may need to support a large number of different encoder types from different vendors or different versions. Similarly, the encoder part at the UE may need to support different decoder types at the base station. In addition, for a given chipset vendor on the UE side and the base station side, there may be multiple versions of the same basic AE model because the AE model is updated through training when more data becomes available. This further increases the number of different encoder models and decoder models that need to be processed when trying to find a matching encoder part and decoder part of the AE.

[0020] Therefore, it is necessary to effectively control the encoding and decoding of uplink control signaling through AE or other machine learning (ML) models. Summary of the invention

[0021] According to an embodiment, a method for controlling wireless communication in a wireless communication network is provided. According to the method, a wireless device receives control data provided by a first node of the wireless communication network. The control data indicates a set of one or more supported ML models for encoding and decoding uplink control signaling from the wireless device to the first node. The wireless device selects one or more ML models from the set. Based on at least one model of the selected one or more ML models, the wireless device encodes uplink control signaling from the wireless device to the first node.

[0022] According to another embodiment, a method of controlling wireless communications in a wireless communication network is provided. According to the method, a node of the wireless communication network provides control data to one or more wireless devices. The control data indicates a set of one or more supported ML models for encoding and decoding uplink control signaling from the wireless device to the node. Using at least one ML model selected from the set, the node decodes uplink control signaling from at least one of the one or more wireless devices.

[0023] According to another embodiment, a wireless device for operating in a wireless communication network is provided. The wireless device is configured to receive control data provided by a first node of the wireless communication network. The control data indicates a set of one or more supported ML models for encoding and decoding uplink control signaling from the wireless device to the first node. In addition, the wireless device is configured to select one or more ML models from the set. In addition, the wireless device is configured to encode uplink control signaling from the wireless device to the first node based on at least one of the one or more selected ML models.

[0024] According to another embodiment, a wireless device for operating in a wireless communication network is provided. The wireless device includes at least one processor and a memory. The memory contains instructions executable by the at least one processor, whereby the wireless device is operable to receive control data provided by a first node of the wireless communication network. The control data indicates a set of one or more supported ML models for encoding and decoding uplink control signaling from the wireless device to the first node. In addition, the memory contains instructions executable by the at least one processor, whereby the wireless device is used to select one or more ML models from the set. In addition, the memory contains instructions executable by the at least one processor, whereby the wireless device is operable to encode uplink control signaling from the wireless device to the first node based on at least one of the one or more selected ML models.

[0025] According to another embodiment, a node for a wireless communication network is provided. The node is configured to provide control data to one or more wireless devices. The control data indicates a set of one or more supported ML models for encoding and decoding uplink control signaling from the wireless device to the node. In addition, the node is configured to decode uplink control signaling from at least one of the one or more wireless devices using at least one ML model selected from the set.

[0026] According to another embodiment, a node for a wireless communication network is provided. The node includes at least one processor and a memory. The memory contains instructions executable by the at least one processor, whereby the node is operable to provide control data to one or more wireless devices. The control data indicates a set of one or more supported ML models for encoding and decoding uplink control signaling from the wireless device to the node. In addition, the memory contains instructions executable by the at least one processor, whereby the node is operable to decode uplink control signaling from at least one of the one or more wireless devices using at least one ML model selected from the set.

[0027] According to another embodiment, a computer program or computer program product is provided, for example, in the form of a non-transitory storage medium, which includes program code to be executed by at least one processor of a wireless device to operate in a wireless communication network. Execution of the program code causes the wireless device to receive control data provided by a first node of the wireless communication network. The control data indicates a set of one or more supported ML models for encoding and decoding uplink control signaling from the wireless device to the first node. In addition, execution of the program code causes the wireless device to select one or more ML models from the set. In addition, execution of the program code causes the wireless device to encode uplink control signaling from the wireless device to the first node based on at least one of the one or more selected ML models.

[0028] According to another embodiment, a computer program or computer program product is provided, for example, in the form of a non-transitory storage medium, comprising program code to be executed by at least one processor of a node for a wireless communication network. Execution of the program code causes the node to provide control data to one or more wireless devices. The control data indicates a set of one or more supported ML models for encoding and decoding uplink control signaling from the wireless device to the node. In addition, execution of the program code causes the node to decode uplink control signaling from at least one of the one or more wireless devices using at least one ML model selected from the set.

[0029] Details of such and further embodiments will be apparent from the following detailed description of the embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 MU-MIMO operation in a wireless communication network is schematically illustrated.

[0031] Figure 2 Schematically illustrating CSI Type II feedback for NR technology.

[0032] Figure 3 The AE architecture is schematically shown.

[0033] Figure 4 A wireless communication network according to an embodiment is schematically illustrated.

[0034] Figure 5 The CSI report according to the embodiment is schematically illustrated.

[0035] Figure 6 An example of a process according to an embodiment is shown.

[0036] Figure 7 Another example of a process according to an embodiment is shown.

[0037] Figure 8 A flow chart is shown for schematically illustrating a method according to an exemplary embodiment.

[0038] Fig. 9 A flow chart is shown for schematically illustrating a further method according to an embodiment.

[0039] Fig.10 The structure of a wireless device according to an embodiment is schematically shown.

[0040] Fig.11 The structure of a network node according to an embodiment is schematically shown.

[0041] Fig.12 The interaction of a host and a wireless device according to an embodiment is schematically illustrated. DETAILED DESCRIPTION

[0042] In the following, the concepts of exemplary embodiments according to the present disclosure will be explained in more detail and with reference to the accompanying drawings. The illustrated embodiments relate to the control of wireless communications in a wireless communication network, and in particular to the control of uplink signaling from a wireless device to a node of the wireless communication network. The wireless communication network may be, for example, a cellular network, for example, as specified by 3GPP. Then, the wireless communication may be, for example, based on NR technology, LTE technology, or future 6G (sixth generation) technology. However, these concepts may also be applied to other types of wireless communication networks, for example, based on WLAN (wireless local area network) technology. As used herein, the term "wireless device" (WD) refers to a device capable of, configured, arranged, and / or operable to communicate wirelessly with a network node and / or other WD. Unless otherwise specified, the term WD may be used interchangeably with UE herein. Wireless communications may involve sending and / or receiving wireless signals using electromagnetic waves, radio waves, infrared waves, and / or other types of signals suitable for transmitting information through the air. In some embodiments, the WD may be configured to send and / or receive information without direct human interaction. For example, the WD may be designed to send information to the network on a predetermined schedule when triggered by an internal or external event or in response to a request from the network. Examples of WD include, but are not limited to, smart phones, mobile phones, cellular phones, voice over IP (VoIP) phones, wireless local loop phones, desktop computers, personal digital assistants (PDAs), wireless cameras, game consoles or devices, music storage devices, playing appliances, wearable terminal devices, wireless endpoints, mobile stations, tablet computers, laptop computers, laptop embedded devices (LEEs), laptop computer installation equipment (LMEs), smart devices, wireless client equipment (CPEs), vehicle-mounted wireless terminal devices, networked vehicles, etc. In some examples, in the Internet of Things (IoT) scenario, WD can also represent a machine or other device that performs monitoring and / or measurement, and transmits the results of such monitoring and / or measurement to another WD and / or network node. WD in this case can be a machine-to-machine (M2M) device, which can be referred to as a machine type communication (MTC) device in the 3GPP context. As a specific example, WD can be a UE that implements the 3GPP narrowband IoT (NB-IoT) standard. Specific examples of such machines or devices are sensors, metering devices such as power meters, industrial machinery, household or personal appliances (e.g., refrigerators, televisions, etc.), or personal wearable devices (e.g., watches, fitness trackers, etc.). In other scenarios, a WD may represent a vehicle or other device capable of monitoring and / or reporting its operating status or other functions associated with its operation. A WD as described above may represent an endpoint of a wireless connection, in which case the device may be referred to as a wireless terminal. In addition, a WD as described above may be mobile, in which case it may also be referred to as a mobile device or a mobile terminal.

[0043] In the illustrated concept, AI (artificial intelligence) based encoding and decoding of uplink control signaling from wireless devices is applied, i.e., an ML model is used to efficiently encode and decode uplink control signaling. The uplink control signaling may, for example, include CSI, beam management information, location information, etc. The ML model may, for example, be based on a combination of Figure 3 The explained two-sided architecture implements AE. Specifically, it is assumed that the ML model is assumed to include an encoder part located at the UE and a decoder part located at a node of the wireless communication network, typically at a base station that receives uplink control signaling. In the concept shown, the base station can provide capability information indicating the ML models supported by the base station, for example, in the form of a list of identifiers (IDs) of the ML models supported by the base station. In the following, this capability of the base station (BS) is also represented as "BS capability information". In some scenarios, the base station may broadcast the BS capability information in the system information of the cell, for example on the PBCH (physical broadcast channel) of the cell. In such a case, the BS capability information can be included, for example, in a MIB (master information block) or a SIB (system information block). However, other types of control data may also be used to transmit BS capability information, for example, messages of switching control signaling or other connection control signaling.

[0044] When the UE receives the BS capability information, it can take into account the ML models supported by the base station when reporting its own capabilities to the base station. In the following, information indicating the capabilities of the UE is also represented as "UE capability information". The UE capability information reported by the UE can identify the ML model supported by the UE, which is selected based on the BS capability information. Specifically, the UE capability information can indicate a selection from the ML models supported by the base station, which is also supported by the UE. Therefore, if the base station does not support the ML model, the UE does not need to report the supported ML model. In this way, the overhead and complexity of capability signaling from the UE to the base station can be reduced. In some cases, the UE capability information reported by the UE can indicate the final selection of the ML model, which will be applied by the UE to encode the uplink control signaling. In other cases, the UE capability information reported by the UE can indicate a pre-selection of the ML model, and the ML model to be applied by the UE can be selected by the base station. This selection of the base station can then be indicated in further control signaling from the base station to the UE.

[0045] When considering the above application scenarios of two-sided AE, the capability signaling of the shown concept is beneficial because it allows to effectively consider the ML model to be two-sided, where the encoder part is at the UE and the decoder part is at the base station, and these two parts are usually trained jointly. Therefore, an ML model with matching parts at the base station and the UE can be selected in an efficient way. Reporting ML models in the UE capability information that cannot be paired with the base station can be avoided. If the BS capability information is broadcast, a single broadcast message is sufficient to inform all UEs in the cell about the ML models supported by the base station. Here, indicating a large set of supported ML models is beneficial because different UEs can support different ML models.

[0046] Therefore, in the concept shown, the network side, in particular the base station, can provide control data, such as broadcast MIB or SIB or handover control message, which indicates a set of network side capabilities that support specific features, in particular, the encoding and decoding of uplink control signaling based on two-sided AI / ML (set A). The UE can then receive the control data, compare the indicated capabilities with its own set of capabilities for the same features (set B), and then report back to the network side the intersection of set A and set B.

[0047] Figure 4 An exemplary structure of a communication network is shown. In the example shown, it is assumed that the communication network is a wireless communication network specified by 3GPP. In particular, Figure 4 A plurality of UEs 10 served by a base station 100 of a wireless communication network are shown. Here, it should be noted that the base stations 100 may each serve a plurality of cells within the coverage area of ​​the wireless communication network. The base stations 100 may, for example, each correspond to a gNB of NR technology or an eNB of LTE technology. The base stations 100 may be considered to be part of the RAN (Radio Access Network) of the wireless communication network and may therefore also be denoted as access nodes. In addition, Figure 4 A CN (core network) 110 of a wireless communication network is schematically shown. Figure 4 In the figure, CN 110 is illustrated as including one or more gateways 120 and one or more control nodes 130. Gateway 120 may be responsible for processing user plane traffic for UE 10, for example, by forwarding user plane data traffic from UE 10 to a network destination or by forwarding user plane data traffic from a network source to UE 10. Here, the network destination may correspond to another UE 10, an internal node of a wireless communication network, or an external node connected to a wireless communication network. Similarly, the network source may correspond to another UE 10, an internal node of a wireless communication network, or an external node connected to a wireless communication network. Control node 130 may be used to control user data traffic, for example, by providing control data to base station 100, gateway 120 and / or UE 10.

[0048] As indicated by the double-headed arrows, the base station 100 may send downlink transmissions to the UE, and the UE may send uplink transmissions to the base station 100. The downlink transmissions and uplink transmissions may be used to provide various types of services to the UE, such as voice services, multimedia services, or data services. Such services may be hosted, for example, by corresponding network nodes in the CN 110. As an example, Figure 4 A service platform 150 is shown provided in the CN 110. In addition, such services may be hosted externally, for example, by an AF (Application Function) connected to the CN 110. As an example, Figure 4 One or more application servers 180 connected to the CN 110 are shown. The application server 180 may be connected to the CN 110, for example, via the Internet or some other wide area communication network. The service platform 150 may be based on a server or cloud computing system and hosted by one or more hosts. Similarly, the (one or more) application servers 180 may be based on a server or cloud computing system and hosted by one or more hosts. The (one or more) application servers 160 may include or be associated with one or more AFs, which enable interaction with the CN 110 to provide one or more services corresponding to one or more applications to the UE 10. These services or applications may generate user plane data traffic transmitted by downlink transmissions and / or uplink transmissions between the base station 100 and the corresponding UE 10. Therefore, the application server 180 may include or correspond to the above-mentioned network destination and / or network source of the user data traffic. In the corresponding UE 10, such services may be based on applications (or simply "apps") executed on the UE 10. Such applications may be pre-installed or installed by the user. Such an application may generate at least a portion of the user plane traffic between the UE 10 and the base station 100 .

[0049] In the concept of the diagram, for Figure 4 At least some of the UEs 10 illustrated in the figure, uplink control signaling to the base station 100 may be based on encoding and decoding using ML models, in particular AE with a dual-sided architecture having an encoder part at the UE 10 and a decoder part at the base station 100. In this context, the capability signaling of the illustrated concept may facilitate the efficient selection of (one or more) ML models to be applied for encoding and decoding of the uplink control signaling. The uplink control signaling may, for example, convey beam management information, CSI and / or location information.

[0050] Figure 5 It schematically shows how to apply the AE selected based on the concept shown to control the UE and the base station (BS) (for example, Figure 41 and one of the base stations 100 shown in the figure) for MIMO operation, such as MU-MIMO operation. For multiple transmit (TX) ports and receive (RX) ports, the input to the encoder on the UE side includes CSI representing the MIMO channel estimated on several subcarriers (sc). For CSI compression, the encoder is implemented in the UE and the decoder is implemented in the base station. Appropriate selection of AE allows efficient compression of the CSI. Therefore, CSI feedback with both high resolution and low uplink overhead can be provided. Similar AE-based compression can also be achieved for other types of uplink control signaling, such as beam management information or position information.

[0051] The list of ML model IDs may be added to the system information broadcast by the base station 100, for example, as part of the cell configuration parameters in the SIB. Based on the list of broadcasted ML model IDs, a UE 10 intending to register to the cell will be able to check whether there is a suitable ML model, i.e., whether the base station supports any ML model that matches the ML model supported by the UE, and select such a suitable (one or more) ML model. Typically, such a matching ML model will include an encoder part and a decoder part that have been jointly trained. The UE 10 may then include the list of selected ML model IDs in the UE capability information reported to the base station 100 during connection establishment. In some cases, if the UE determines that the base station 100 supports one or more ML models that the UE 10 does not currently support, the UE 10 may download the model from the network server before the connection is established. In some cases, the UE 10 may report one model ID per type of uplink control signaling to be encoded / decoded, for example, one ML model for CSI, one ML model for beam management, one ML model for location information, etc. Therefore, the UE 10 may also apply multiple ML models in parallel. In some cases, the UE 10 may also report some or more of the ML model IDs that are not supported by the UE 10. Thus, the UE 10's support for ML models may be indicated in the form of a positive selection of supported ML models and / or a negative selection of unsupported ML models.

[0052] In the following, the illustrated concepts will be explained in more detail by referring to an exemplary process involving ML-based encoding and decoding of uplink control signaling. In these examples, it is assumed that the encoder and decoder parts of multiple ML models have been jointly trained, for example using a bilateral development domain setup between a base station and a UE or chipset vendor. A corresponding ID is assigned to each pair of such trained ML models. For example, an AE having an encoder part and a decoder part may have a UE model ID and a BS model ID, or a model pair ID. In some scenarios, it may also occur that a BS model ID can be paired with multiple different UE model IDs, that is, it performs well with multiple different UE model IDs, or vice versa. Therefore, a "model ID" may also be summarized as a set of model IDs, or may represent a set of models including a "version ID", which identifies different versions of the same model, for example, a version obtained by further training updates. If a particular ML model pair on the UE side and the base station side provides good performance in encoding and decoding uplink control signaling, the ML model pair may be classified as a "match". As described above, such a match may be the result of joint training of ML models. However, such good performance may also occur for ML model pairs that are not jointly trained. As described above, initially the base station may broadcast or otherwise provide BS capability information indicating the ML models supported by the base station to (multiple) UEs, e.g., in the form of a list of model IDs. Based on this information, the UE may decide which ML models to report as supported (or not supported) in the UE capability information sent to the base station, e.g., during connection establishment. Thus, the UE does not need to report model IDs for matching ML models that it supports but the base station does not support.

[0053] Furthermore, the UE may use the BS capability information to trigger the download of certain ML models: if the BS capability information indicates that the base station supports a certain model ID, the UE may download the corresponding ML model from the server. Such downloading may be performed without explicit knowledge of the base station or CN of the wireless communication network, i.e. "over the top". When the UE has installed a new ML model, it may send a report to the base station that it supports the new model ID, which will then also be supported by the base station.

[0054] Since the download and installation of the new ML model may take some time, the UE may have reported the currently supported ML model to the base station. Later, after downloading and installing the new ML model, the UE may request the base station to schedule the UE with uplink transmission to allow the UE to update its capabilities, i.e., to inform the base station that it now supports the new model ID through new UE capability information. The base station may then choose to reconfigure the uplink control signaling from the UE to use the new ML model.

[0055] In other scenarios, the UE may report to the base station that it does not currently support a certain ML model (identified by the model ID), but will start downloading it for later use. There may be some maximum time that such a download is allowed. For example, the network may assume that the ML model has been downloaded after a certain time interval. The time interval may be predefined or configurable. After the time interval expires, the base station may then attempt to configure the UE with the newly installed ML model for uplink control signaling.

[0056] There may also be a case of a "blank" UE, i.e., when the UE has not downloaded an ML model and first reads the BS capability information, then downloads a matching ML model, and then reports UE capability information indicating support for the matching ML model to the base station.

[0057] In some cases, the BS capability information indicating the supported ML models may also be supplemented with additional information, such as an indication of the frequency bands or frequency ranges in which a certain model ID is supported. For example, a base station may support the use of ML-based encoding and decoding or the use of a certain ML model only in certain frequency bands, and such selective support may be indicated in the BS capability information.

[0058] In some cases, the BS capability information may also indicate that the base station does not support ML-based encoding and decoding, so that the UE does not need to report the relevant capabilities in the UE capability information. In some cases, the base station may indicate such limitation of support for ML-based encoding and decoding for certain types of uplink control signaling, for example, by indicating that ML-based encoding and decoding is not supported for CSI, is not supported for beam management information, or is not supported for location information.

[0059] Figure 6 Schematically shows an example of a process of using the capability signaling concept shown when establishing a connection between a UE 10 and a base station 100. The UE 10 and the base station 100 may correspond to, for example, Figure 4 One of the base stations 100 and one of the UEs 10 are shown.

[0060] exist Figure 6In the example of , the base station 100 broadcasts system information 601. The system information includes a model ID list indicating the ML models supported by the base station 100. The model ID list specifically indicates the decoders supported on the base station side. The model ID list may indicate the ML models supported for each type of uplink control signaling, for example, whether the ML model is supported for CSI, beam management, and location information. In some cases, the system information 601 may include such a model ID list per frequency band, per frequency band group, or per frequency range. (Multiple) model ID lists may be included in an existing SIB, for example, in a corresponding information element added to an existing SIB. Optionally, the model ID list may be included in a dedicated SIB, for example, for indicating ML-related information.

[0061] Assuming that the UE 10 decides to camp on a cell served by the base station 100, the UE 10 receives the system information 601 and reads the included model ID list (s). Based on this information, the UE 10 can check whether it can support one or more of the indicated ML models, i.e., select the corresponding ML model supported by the UE 10, as indicated by block 602. This selection can be done for each type of uplink control signaling. In some cases, the UE 10 can also select an ML model that the UE 10 does not yet support and trigger the download of the ML model from the server, as indicated by block 603.

[0062] When initiating connection establishment with the base station 100, the UE 10 then sends a report 604 of the selected ML model to the base station 100. For example, the report 604 may identify the ML models supported by the UE 10 (including the ML models optionally downloaded at box 603) in the form of one or more model ID lists. The model ID list may indicate the ML models supported for each type of uplink control signaling, for example, whether the ML model is supported for CSI, for beam management, for location information. In some cases, the report 604 may include such a model ID list per frequency band, per frequency band group, or per frequency range. The report 604 may, for example, correspond to or be part of a UE capability report message. It should be noted that it may happen that the UE 10 does not support ML-based encoding and decoding for all types of uplink control signaling supported by the base station, and such limitations may also be indicated in the report 604.

[0063] As further shown, the base station 100 sends an RRC (Radio Resource Control) connection setup message 605 to the UE 10, and the UE responds with an RRC connection setup complete message 606 to configure the connection between the UE 10 and the base station 100. The RRC connection setup message 605 may indicate the ML model(s) that the UE 10 is to apply when sending uplink control signaling 607 to the base station 100. Alternatively, the UE 10 may indicate in the RRC connection setup message 606 which of the supported ML models it will apply when sending uplink control signaling 607.

[0064] It should be noted that Figure 6 The process may include further messages which are not shown for the sake of a better overview.

[0065] Figure 7 Schematically shows an example of a process of using the capability signaling concept shown in a handover (HO) of a UE 10 from a source base station 100' to a target base station 100. The UE 10 and the base stations 100, 100' may correspond to, for example, Figure 4 One of the base station 100 and UE 10 shown. Figure 7 In the handover scenario shown, a list of ML model IDs supported by the target cell (corresponding to the target base station 100) is provided to the UE 10 during the handover process. In particular, the source cell (corresponding to the source base station 100') and the target cell may exchange information about the ML models supported by the target cell.

[0066] exist Figure 7 In the example of , the handover of UE 10 is initiated by a handover request (Handover Request) 701 from the source base station 100' to the target base station 100. The target base station 100 confirms the requested handover by sending a handover request confirmation (HORequest ACK) 702 to the source base station 100'. The handover request confirmation 702 includes a model ID list indicating the ML models supported by the target base station 100. The model ID list may indicate the ML models supported for each type of uplink control signaling, for example, whether the ML model is supported for CSI, for beam management, and for location information. In some cases, the handover request confirmation 702 may include such a model ID list per frequency band, per frequency band group, or per frequency range. The model ID list may be included in a corresponding information element of the handover request confirmation 702.

[0067] The source base station 100' then forwards the model ID list to the UE 10 in an RRC Connection Reconfiguration message 703. The UE 10 receives the RRC Connection Reconfiguration message 703 and reads the included model ID list. Based on this information, the UE 10 can check whether it supports one or more of the indicated ML models, i.e., select the corresponding ML model supported by the UE 10, as indicated by box 704. This selection can be done for each type of uplink control signaling. In some cases, the UE 10 can also select an ML model that the UE 10 does not yet support and trigger the download of the ML model from the server, as indicated by box 705.

[0068] Then, the UE 10 sends an RRC connection reconfiguration complete message 706 to the target base station 100. The RRC connection reconfiguration complete message 706 indicates the selected ML model of box 704, for example in the corresponding information element of the RRC connection reconfiguration complete message 706. The RRC connection reconfiguration complete message 706 can identify the ML models supported by the UE 10 (including the ML models optionally downloaded at box 603) in the form of one or more model ID lists. The model ID list can indicate the ML models supported for each type of uplink control signaling, for example, whether the ML model is supported for CSI, for beam management, and for location information. In some cases, the RRC connection reconfiguration complete message 706 may include such a model ID list per frequency band, per frequency band group, or per frequency range. It should be noted that it may happen that the UE 10 does not support ML-based encoding and decoding for all types of uplink control signaling supported by the base station, and such restrictions may also be indicated in the RRC connection reconfiguration complete message 706.

[0069] As further shown, the UE 10 may then send uplink control signaling 707 based on at least one of the supported ML models. Figure 7 The process may include additional messages, which are not shown for better overview. For example, such messages may be used to further negotiate the selection of one or more models from the set of ML models indicated as supported by UE 10 in RRC connection reconfiguration complete message 706.

[0070] Figure 8 A flow chart is shown to illustrate a method that can be used to implement the concepts shown. Figure 8The method may be used to implement the described concepts in a wireless device (eg corresponding to one of the above-described UEs 10) operating in a wireless communication network.

[0071] If the wireless device uses a processor-based implementation, then Fig. 9 At least some of the steps of the method may be executed and / or controlled by one or more processors of the wireless device. Such a wireless device may also include a memory storing program code, the program code being used to implement the following description Fig. 9 at least some of the functions or steps of the method.

[0072] At step 910, the node provides control data to one or more wireless devices, e.g., corresponding to one or more of the UEs 10 described above. The control data indicates a set of one or more supported ML models for encoding and decoding uplink control signaling from the wireless device to the node. The control data may indicate the supported ML models, e.g., in the form of a list of model IDs. The node may send the control data in a wireless transmission. The wireless transmission may be a broadcast transmission, e.g., such as for a wireless device 100. Figure 6 In some scenarios, the wireless device may receive control data in a wireless transmission from a second node of the wireless communication network. For example, this may be the case in a handover scenario. The control data may then control the handover of the wireless device from the second node to the first node.

[0073] The uplink control signaling may include CSI. Additionally or alternatively, the uplink control signaling may include beam management information. Additionally or alternatively, the uplink control signaling may include location information. The ML model may be neural network based and may correspond to an AE with a bilateral architecture having an encoder portion at the wireless device and a decoder portion at the base station, e.g., Figure 3 shown.

[0074] The wireless device selects one or more ML models from the set at step 820. The selection may be based on a match between the ML models in the set and the ML models supported by the wireless device.

[0075] At step 830, in response to the selection at step 820, the wireless device may download at least one of the ML models selected at step 820. This may be done, for example, in response to the control data indicating that the base station supports the ML model, but the ML model is not yet supported by the wireless device. The wireless device may download at least one ML model from a server accessible via the wireless communication network.

[0076] At step 840, the wireless device may report the selected one or more ML models to the first node. In some scenarios, the wireless device may select multiple ML models from the set and report the selected multiple ML models to the first node. The reporting of step 840 may be accomplished, for example, via a capability information message sent during connection establishment between the wireless device and the base station, such as for example, Figure 6 In addition, the reporting of step 840 may be accomplished by a message of the switching process, for example, as for Figure 7 As explained in the RRC connection reconfiguration complete message 706.

[0077] At step 850, the wireless device may receive additional control data. If at steps 830 and 840, the wireless device selected a plurality of ML models from the set and reported the selected plurality of ML models to the first node, the additional control data may indicate at least one of the reported ML models to be applied by the wireless device to encode uplink control signaling. Such additional control data may be, for example, Figure 6 The RRC connection establishment message 605 is transmitted or transmitted by Figure 7 The other message after the RRC connection reconfiguration message 706 is transmitted.

[0078] At step 860, the wireless device encodes uplink control signaling from the wireless device to the first node. The encoding may be done based on at least one of the one or more ML models selected at step 820 and optionally downloaded at step 830.

[0079] Fig. 9 A flow chart is shown to illustrate a method that can be used to implement the concepts shown. Fig. 9 The method may be used to implement the illustrated concept in a node (eg corresponding to one of the base stations 100 described above) of a wireless communication network.

[0080] If the node uses a processor-based implementation, then Fig. 9 At least some of the steps of the method may be executed and / or controlled by one or more processors of the node. Such a node may also include a memory storing program code, the program code being used to implement the following description Fig. 9 at least some of the functions or steps of the method.

[0081] At step 910, the wireless device receives control data provided by a first node of the wireless communication network. The control data indicates a set of one or more supported ML models for encoding and decoding uplink control signaling from the wireless device to the first node. The control data may indicate the supported ML models, for example, in the form of a list of model IDs. The wireless device may receive the control data in a wireless transmission from the first node. The wireless transmission from the first node may be a broadcast transmission, for example, such as a broadcast transmission for a wireless device. Figure 6 In some scenarios, a node may send control data via another node of a wireless communication network. For example, this may be the case in a handover scenario. The control data may then control the handover of a wireless device from another node to the node.

[0082] The uplink control signaling may include CSI. Additionally or alternatively, the uplink control signaling may include beam management information. Additionally or alternatively, the uplink control signaling may include location information. The ML model may be neural network based and may correspond to an AE with a bilateral architecture having an encoder portion at the wireless device and a decoder portion at the base station, e.g., Figure 3 shown.

[0083] At step 920, the node may receive a report from at least one of the wireless devices. The report may indicate one or more ML models selected from the set. In some scenarios, the report may indicate multiple selected ML models from the set. The report of step 920 may be received, for example, in a capability information message sent during connection establishment between the wireless device and the base station, such as for example, Figure 6 In addition, the report of step 920 may be received in a message of the handover process, for example, as for Figure 7 As explained in the RRC connection reconfiguration complete message 706.

[0084] At step 930, the node may provide additional control data to at least one of the one or more wireless devices. If at step 930, the wireless device reported multiple ML models selected from the set, the additional control data may indicate at least one of the reported ML models to be applied by the wireless device to encode uplink control signaling. Such additional control data may be provided, for example, by Figure 6 The RRC connection establishment message 605 is transmitted or transmitted by Figure 7 The other message after the RRC connection reconfiguration message 706 is transmitted.

[0085] At step 940 , the node decodes uplink control signaling from at least one wireless device. The decoding may be done based on at least one of the one or more ML models indicated by the control data of step 910 .

[0086] Fig.10 Schematically illustrates a processor-based implementation of a wireless device 1000 for operating in a wireless communication network, which can be used to implement the above concepts. Fig.10 The illustrated structure may be used to implement the concepts in one or more of the UEs 10 described above.

[0087] As shown in the figure, the wireless device 1000 may include a wireless interface 1010. The wireless interface 1010 may be used to perform wireless communications with one or more nodes (such as the base station 100 described above) of a wireless communication network.

[0088] In addition, the wireless device 1000 may include one or more processors 1050 coupled to the wireless interface 1010 and a memory 1060 coupled to the (multiple) processors 1050. For example, the wireless interface 1010, the processor 1050 and the memory 1060 may be coupled via one or more internal bus systems of the wireless device 1000. The memory 1060 may include a read-only memory (ROM) such as a flash ROM, a random access memory (RAM) such as a dynamic RAM (DRAM) or a static RAM (SRAM), a large-capacity memory such as a hard disk or a solid-state disk, etc. As shown, the memory 1060 may include software 1070 and / or firmware 1080. The memory 1060 may include appropriately configured program code to be executed by the processor 1050 in order to implement the above-mentioned functions for controlling wireless communication, such as in conjunction with Figure 8 Explained.

[0089] It should be understood that Fig.10 The structure shown in is only schematic, and the wireless device 1000 may actually include additional components, such as additional interfaces or additional processors, which are not shown for clarity. In addition, it should be understood that the memory 1060 may include additional program code for implementing known functions of a UE supporting NR technology or LTE technology. According to some embodiments, a computer program for implementing the functions of the wireless device 1000 may also be provided, for example, in the form of a physical medium storing program code and / or other data to be stored in the memory 1060, or by making the program code downloadable or by streaming.

[0090] Fig.11 Schematically illustrates a processor-based implementation of a node 1100 for a wireless communication network, which can be used to implement the above concepts. Fig.11The illustrated structure may be used to implement the concepts in one or more of the above-described base stations 100 or similar access nodes.

[0091] As shown, node 1100 may include a wireless interface 1110 and a network interface 1120. Wireless interface 1110 may be used to wirelessly communicate with one or more wireless devices such as the above-mentioned UE 10. Network interface 1120 may be used to communicate with one or more other nodes of a wireless communication network such as other access nodes or CN nodes.

[0092] In addition, the node 1100 may include one or more processors 1150 coupled to the interfaces 1110, 1120 and a memory 1160 coupled to the (multiple) processors 1150. For example, the interfaces 1110, 1120, the processor 1150 and the memory 1160 may be coupled via one or more internal bus systems of the node 1100. The memory 1160 may include a ROM such as a flash ROM, a RAM such as a DRAM or SRAM, a large-capacity memory such as a hard disk or a solid-state disk, etc. As shown, the memory 1160 may include software 1170 and / or firmware 1180. The memory 1160 may include appropriately configured program code to be executed by the processor 1150 in order to implement the above-mentioned functions for controlling wireless communication, such as in conjunction with Fig. 9 Explained.

[0093] It should be understood that Fig.11 The structure shown in is only schematic, and the node 1100 may actually include additional components, such as additional interfaces or additional processors, which are not shown for clarity. Moreover, it should be understood that the memory 1160 may include additional program code for implementing the known functions of a gNB for NR technology, an eNB for LTE technology, or a similar type of access node. According to some embodiments, a computer program for implementing the functions of the node 1100 may also be provided, for example, in the form of a physical medium storing program code and / or other data to be stored in the memory 1160, or by making the program code downloadable or by streaming.

[0094] Fig.12 A communication diagram of a host 1202 communicating with a UE 1206 via a network node 1204 over a partially wireless connection according to some embodiments. Fig.12 Example implementations of a UE (such as one of the UEs 10 described above), a network node (such as one of the base stations described above), and a host (such as the service platform 150 or application server(s) 180 described above) according to various embodiments are described.

[0095] An embodiment of the host 1202 includes hardware, such as a communication interface, a processing circuit, and a memory. The host 1202 also includes software that is stored in or accessible by the host 1202 and that is executable by the processing circuit. The software includes a host application that is operable to provide services to a remote user, such as a UE 1206, that is connected via an over-the-top (OTT) connection 1250 extending between the UE 1206 and the host 1202. In providing services to the remote user, the host application can provide user data sent using the OTT connection 1250.

[0096] The network node 1204 includes hardware that enables it to communicate with the host 1202 and the UE 1206. The connection 1260 can be direct or through a core network (such as Figure 4 The core network 110 of the present invention) and / or one or more other intermediate networks, such as one or more public, private or managed networks. For example, the intermediate network can be a backbone network or the Internet.

[0097] UE 1206 includes hardware and software, which is stored in UE 1206 or accessible by UE 1206, and can be executed by the processing circuit of UE. The software includes a client application, such as a web browser or an operator-specific "app", which is operable to provide services to human or non-human users via UE 1206 with the support of host 1202. In host 1202, the executing host application can communicate with the executing client application via an OTT connection 1250 terminated at UE 1206 and host 1202. When providing services to the user, the client application of the UE can receive request data from the host application of the host, and provide user data in response to the request data. OTT connection 1250 can transmit both request data and user data. The client application of the UE can interact with the user to generate user data that it provides to the host application via OTT connection 1250.

[0098] The OTT connection 1250 may extend via a connection 1260 between the host 1202 and the network node 1204 and via a wireless connection 1270 between the network node 1204 and the UE 1206 to provide connectivity between the host 1202 and the UE 1206. The connection 1260 and the wireless connection 1270 through which the OTT connection 1250 may be provided have been drawn abstractly to illustrate communications between the host 1202 and the UE 1206 via the network node 1204, without explicitly involving any intermediate devices and the precise routing of messages via these devices.

[0099] As an example of transmitting data via the OTT connection 1250, in step 1208, the host 1202 provides user data, which can be executed by executing a host application. In some embodiments, the user data is associated with a specific human user interacting with the UE 1206. In other embodiments, the user data is associated with the UE 1206, and the UE 1206 shares the data with the host 1202 without explicit human interaction. In step 1210, the host 1202 initiates a transmission carrying the user data to the UE 1206. The host 1202 may initiate the transmission in response to a request sent by the UE 1206. The request may be caused by human interaction with the UE 1206 or by the operation of a client application executed on the UE 1206. According to the teachings of the embodiments described throughout the present disclosure, the transmission may be delivered via the network node 1204. Therefore, according to the teachings of the embodiments described throughout the present disclosure, in step 1212, the network node 1204 sends the user data carried in the transmission initiated by the host 1202 to the UE 1206. In step 1214 , UE 1206 receives the user data carried in the transmission, which may be performed by a client application executing on UE 1206 associated with a host application executed by host 1202 .

[0100] In some examples, UE 1206 executes a client application that provides user data to host 1202. The user data may be provided as a reaction or response to data received from host 1202. Therefore, in step 1216, UE 1206 may provide the user data, which may be performed by executing the client application. In providing the user data, the client application may further consider user input received from the user via the input / output interface of UE 1206. Regardless of the specific manner in which the user data is provided, UE 1206 initiates transmission of the user data to host 1202 via network node 1204 in step 1218. In step 1220, in accordance with the teachings of the embodiments described throughout the present disclosure, network node 1204 receives the user data from UE 1206 and initiates transmission of the received user data to host 1202. In step 1222, host 1202 receives the user data carried in the transmission initiated by UE 1206.

[0101] The illustrated concepts may help improve the performance of OTT services provided to UE 1206 using OTT connection 1250, in which wireless connection 1270 forms the backend. More precisely, the teachings of these embodiments may improve the efficiency of uplink control signaling and thereby allow more precise and efficient control of data transmission on the backend of OTT connection 1250.

[0102] In an example scenario, plant status information may be collected and analyzed by host 1202. As another example, host 1202 may process audio and video data that may have been retrieved from the UE for use in creating a map. As another example, host 1202 may collect and analyze real-time data to assist in controlling vehicle congestion (e.g., controlling traffic lights). As another example, host 1202 may store surveillance videos uploaded by the UE. As another example, host 1202 may store or control access to media content such as video, audio, VR, or AR that may be broadcast, multicast, or unicast to the UE. As other examples, host 1202 may be used for energy pricing, remote control of non-time-critical electrical loads to balance power generation demand, location services, presentation services (such as compiling charts based on data collected from remote devices, etc.), or any other function for collecting, retrieving, storing, analyzing, and / or transmitting data.

[0103] In some examples, a measurement process may be provided for the purpose of monitoring data rates, latency, and other factors that are improved by one or more embodiments. There may also be an optional network function for reconfiguring the OTT connection 1250 between the host 1202 and the UE 1206 in response to changes in the measurement results. The measurement process and / or the network function for reconfiguring the OTT connection may be implemented in software and hardware of the host 1202 and / or the UE 1206. In some embodiments, sensors (not shown) may be deployed in or associated with other devices through which the OTT connection 1250 passes; the sensors may participate in the measurement process by providing the values ​​of the monitoring quantities illustrated above, or providing the values ​​of other physical quantities from which the software can calculate or estimate the monitoring quantities. The reconfiguration of the OTT connection 1250 may include message formats, retransmission settings, preferred routes, etc.; the reconfiguration does not require direct changes to the operation of the network node 1204. Such procedures and functions may be known and implemented in the art. In some embodiments, the measurement may involve proprietary UE signaling, which facilitates the host 1202 to measure throughput, propagation time, latency, etc. The measurements may be made by software using the OTT connection 1250 to cause messages to be transmitted, particularly empty or "dummy" messages, while monitoring propagation times, errors, etc.

[0104] It can be seen that the concepts described above can be used to effectively manage the use of ML-based encoding and decoding of uplink control signaling, for example by AE. In particular, through the capability signaling of the shown concepts, the use of ML-based encoding and decoding can be managed with low overhead and complexity cost.

[0105] In the illustrated concept, the use of AE-based uplink control signaling can provide several benefits: if the uplink control signaling conveys CSI, the AE can effectively include a nonlinear transformation, which improves compression performance and thus improves MU-MI MO performance for the same uplink overhead. In addition, the AE can be trained to exploit long-term redundancy in the propagation environment and / or site, such as antenna configuration, for compression purposes. For example, a particular AE does not need to work well for all possible deployments. By learning which channel inputs need (and / or which do not need) to be reliably reconstructed at the base station, improved compression performance can be obtained. Further, the AE can be trained to compensate for antenna array irregularities; for example, non-uniformly spaced antenna elements and non-half-wavelength element spacing. In addition, the AE can be updated to compensate for hardware that fails as the product ages, for example through transfer learning and training.

[0106] It should be understood that the embodiments and implementation schemes explained above are merely illustrative and are susceptible to various modifications. For example, the concepts shown can be applied in conjunction with various communication technologies, without limitation to wireless technologies or technologies specified by 3GPP. In addition, the concepts shown can be applied to ML-based encoding and decoding of various signaling, without limitation to the above-mentioned examples of CSI, beam management information, or location information transmitted by uplink control signaling. In addition, it should be understood that the above-mentioned concepts can be implemented using correspondingly designed software executed by one or more processors of an existing device or apparatus, or by using dedicated device hardware. In addition, it should be noted that the nodes, devices, or apparatus shown can each be implemented as a single device or a system of multiple interacting devices or modules, for example, a virtualized cloud component.

Claims

1. A method for controlling wireless communication in a wireless communication network, the method comprising: The wireless device (10; 1000; 1206) receives control data provided by a first node (100; 1100; 1204) of the wireless communication network, the control data indicating a set of one or more supported machine learning (ML) models for encoding and decoding uplink control signaling from the wireless device (10; 1000; 1206) to the first node (100; 1100; 1204); The wireless device (10; 1000; 1206) selects one or more ML models from the set; and Based on at least one of the selected one or more ML models, the wireless device (10; 1000; 1206) encodes uplink control signaling from the wireless device (10; 1000; 1206) to the first node (100; 1100; 1204).

2. The method according to claim 1, comprising: The wireless device (10; 1000; 1206) reports the selected one or more ML models to the first node (100; 1100; 1204).

3. The method according to claim 2, comprising: The wireless device (10; 1000; 1206) selects a plurality of ML models from the set and reports the ML models to the first node (100; 1100; 1204); and The wireless device (10; 1000; 1206) receives further control data from the first node (100; 1100; 1204), the further control data indicating at least one of the reported ML models to be applied by the wireless device (10; 1000; 1206) to encode the uplink control signaling.

4. The method according to any one of claims 1 to 3, wherein the wireless device (10; 1000; 1206) receives control data in a wireless transmission from the first node (100; 1100; 1204).

5. The method of claim 4, wherein the wireless transmission from the first node (100; 1100; 1204) is a broadcast transmission.

6. The method according to any one of claims 1 to 3, wherein the wireless device (10; 1000; 1206) receives the control data in a wireless transmission from a second node (100') of the wireless communication network.

7. The method according to claim 6, wherein the control data controls a handover of the wireless device (10; 1000; 1206) from the second node (100') to the first node (100).

8. The method according to any one of claims 1 to 7, comprising: In response to selecting the one or more ML models from the set, the wireless device (10; 1000; 1206) Download at least one of the one or more selected ML models.

9. The method according to any one of claims 1 to 8, wherein: The uplink control signaling includes channel state information.

10. The method according to any one of claims 1 to 9, wherein: The uplink control signaling includes beam management information.

11. The method according to any one of claims 1 to 10, wherein: The uplink control signaling includes location information.

12. The method according to any one of claims 1 to 11, wherein: The ML model is based on a neural network.

13. A method for controlling wireless communications in a wireless communications network, the method comprising: A node (100; 1100; 1204) of the wireless communication network provides control data to one or more wireless devices (10; 1000; 1206), the control data indicating a set of one or more supported machine learning (ML) models for encoding and decoding uplink control signaling from the wireless device (10; 1000; 1206) to the node (100; 1100; 1204); and Using at least one ML model selected from the set, the node decodes uplink control signaling from at least one of the one or more wireless devices (10; 1000; 1206).

14. The method according to claim 13, comprising: The node (100; 1100; 1204) receives a report from the at least one wireless device (10; 1000; 1206), the report indicating the at least one selected ML model.

15. The method according to claim 13 or 14, comprising: The node (100; 1100; 1204) receives a report from the at least one wireless device (10; 1000; 1206), the report indicating a plurality of ML models from the list; The node (100; 1100; 1204) selects at least one ML model from the reported ML models; and The node provides further control data to the at least one wireless device (10; 1000; 1206), the further control data indicating at least one selected one of the reported ML models to be applied by the wireless device (10; 1000; 1206) to encode the uplink control signaling.

16. The method according to any one of claims 13 to 15, wherein the node (100; 1100; 1204) sends the control data in a wireless transmission.

17. The method of claim 16, wherein the wireless transmission is a broadcast transmission.

18. The method according to any one of claims 13 to 15, wherein: The node (100; 1100; 1204) provides the control data via another node (100') of the wireless communication network.

19. The method according to claim 18, wherein the control data controls a handover of the wireless device (10; 1000; 1206) from the further node (100') to the node (100; 1100; 1204).

20. The method according to any one of claims 13 to 19, wherein: The uplink control signaling includes channel state information.

21. The method according to any one of claims 13 to 20, wherein: The uplink control signaling includes beam management information.

22. The method according to any one of claims 13 to 21, wherein: The uplink control signaling includes location information.

23. The method according to any one of claims 13 to 22, wherein: The ML model is based on a neural network.

24. A wireless device (10; 1000; 1206) for operating in a wireless communication network, the wireless device (10; 1000; 1206) being configured to: receiving control data provided by a first node (100; 1100; 1204) of the wireless communication network, the control data indicating a set of one or more supported machine learning (ML) models for encoding and decoding uplink control signaling from the wireless device (10; 1000; 1206) to the first node (100; 1100; 1204); selecting one or more ML models from the collection; and Uplink control signaling from the wireless device (10; 1000; 1206) to the first node (100; 1100; 1204) is encoded based on at least one of the selected one or more ML models.

25. The wireless device (10; 1000; 1206) according to claim 24, wherein the wireless device (10; 1000; 1206) is configured to perform the method according to any one of claims 2 to 12.

26. The wireless device (10; 1000; 1206) according to claim 24 or 25, comprising: at least one processor (1050), and A memory (1060) containing program code executable by the at least one processor (1050), whereby the at least one processor (1050) executing the program code causes the wireless device (10; 1000; 1206) to perform a method according to any one of claims 1 to 12.

27. A node (100; 1100; 1204) for a wireless communication network, the node (100; 1100; 1204) being configured to: providing control data to one or more wireless devices (10; 1000; 1206), the control data indicating a set of one or more supported machine learning (ML) models for encoding and decoding uplink control signaling from the wireless devices (10; 1000; 1206) to the node (100; 1100; 1204); and Uplink control signaling from at least one of the one or more wireless devices (10; 1000; 1206) is decoded using at least one ML model selected from the set.

28. The node (100; 1100; 1204) according to claim 27, wherein the node (100; 1100; 1204) is configured to perform the method according to any one of claims 14 to 23.

29. The node (100; 1100; 1204) according to claim 27 or 28, comprising: at least one processor (1150), and A memory (1160) containing program code executable by the at least one processor (1150), whereby execution of the program code by the at least one processor (1150) causes the node (100; 1100; 1204) to perform a method according to any one of claims 13 to 23.

30. A computer program or computer program product comprising program code to be executed by at least one processor (1050) of a wireless device (10; 1000; 1206) operating in a wireless communication network, whereby execution of the program code causes the wireless device (10; 1000; 1206) to perform a method according to any one of claims 1 to 12.

31. A computer program or a computer program product comprising program code to be executed by at least one processor (1150) of a node (100; 1100; 1204) of a wireless communication network, whereby execution of the program code causes the node (100; 1100; 1204) to perform a method according to any one of claims 13 to 23.

Images