Method for reporting indication of invalid inference / prediction report

Abstract

A method, system and apparatus are disclosed A method implemented in a user equipment configured to communicate with a network node includes: generating (S204), based on a machine learning, ML, model / functionality, a report for one or both of an inference report and a prediction report, the report comprising an indication that the one or both of the inference report and the prediction report is one or more of invalid, out of range, and inaccurate; and transmitting (S206) the report to the network node.

Description

[0001] METHOD FOR REPORTING INDICATION OF INVALID INFERENCE / PREDICTION REPORT

[0002] FIELD

[0003] The present disclosure relates to wireless communications, and in particular, to indicating an invalid inference / prediction report.

[0004] BACKGROUND

[0005] The Third Generation Partnership Project (3GPP) has developed and is developing standards for Fourth Generation (4G) (also referred to as Long Term Evolution (LTE)) and Fifth Generation (5G) (also referred to as New Radio (NR)) wireless communication systems. Such systems provide, among other features, broadband communication between network nodes, such as base stations, and mobile user equipments (UEs), as well as communication between network nodes and between UEs. The 3GPP is also developing standards for Sixth Generation (6G) wireless communication networks.

[0006] Artificial Intelligence (Al) / Machine Learning (ML) for Physical layer (PHY) Artificial Intelligence (Al) and Machine Learning (ML) (AI / ML) have been investigated, both in academia and industry, as promising tools to optimize the design of the air-interface in wireless communication networks. Example use cases include using autoencoders for Channel State Information (CSI) compression to reduce the feedback overhead and improve channel prediction accuracy; using deep neural networks for classifying Line-of-Sight (LOS) and Non-LOS (NLOS) conditions to enhance the positioning accuracy; and using reinforcement learning for beam selection at the network side and / or the User Equipment (UE) side to reduce the signaling overhead and beam alignment latency; and using deep reinforcement learning to learn an optimal precoding policy for complex Multiple Input Multiple Output (MIMO) precoding problems.

[0007] In 3rd Generation Partnership Project (3GPP) New Radio (NR) standardization work, a new release 18 (Rel-18) study item on AI / ML for the NR air interface started in May 2022. This study item explored the benefits of augmenting the air-interface with features enabling improved support of AI / ML based algorithms for enhanced performance and / or reduced complexity / overhead. Through studying a few selected use cases (CSI feedback, beam management, and positioning), this study item aims at laying the foundation for future air-interface use cases leveraging AI / ML techniques. The analysis carried out during the Rel-18 is now considered in the context of Release 19 (Rel-19) work item, e.g., as described in 3GPP Technical Release (TR) 38.843 vl8.0.0. Additionally,during Rel-19, a new study item addressing AI / ML for mobility has been approved. In the context of this new study item, 3GPP will investigate methods for cell-level measurement predictions, and mobility event predictions (e.g. radio link failure (RLF), handover failure, mobility-related events predictions such as A3 / A5).

[0008] Applicability reporting

[0009] The use cases considered in beam prediction that will be standardized as part of 3 GPP Rel-19 work item include spatial beam prediction, and temporal beam prediction. The core idea of AI / ML applied to the RAN is to enable a UE to predict / infer certain performances or certain measurements on a given set A of resources based on experienced performances or performed measurements on a set B of resources, where the resources could be, for example, associated to reference signals (e.g., synchronization signal block (SSB) / CSI-reference signal (RS) or positioning reference signal (PRS) resources) or frequencies, depending on the specific AI / ML use case.

[0010] For example, in the case of AI / ML-based beam management (which is considered by 3GPP in the context of Rel-19), the use case is to predict / infer the “best” beam (or beams) from a Set A of beams (SSB / CSLRS resources) using measurement results from another SetB of beams (SSB / CSLRS resources). In particular, according to, e.g., 3GPP TR 38.843 vl8.0.0, the spatial-domain beam prediction for Set A of beams is based on measurement results of Set B of beams, whereas the temporal beam prediction for Set A of beams is based on the historic measurement results of Set B of beams.

[0011] Whether the UE can perform the AI / ML inference, given a certain Set A / B of resources, depends on whether the UE has available an AI / ML model / functionality that can perform the inference / prediction on those Set A of resources given the measurements on the Set B. In other words, the UE can perform the AI / ML inference depending on whether the AI / ML model / functionality is applicable under that Set A / B configuration. For this reason, the network may need to know information related to whether an AI / ML model / functionality is applicable at the UE so the network can properly configure the UE with an inference configuration that enables the AI / ML model / functionality to potentially outperform conventional non-AI / ML based schemes.

[0012] Accordingly, during the 3GPP Rel-18 study item, applicability reporting has been discussed to allow the UE to inform the network node about the applicability of an AI / ML model / functionality while the UE is connected to this network node. An AI / ML model / functionality may be applicable depending on a number of factors (i.e.

[0013] “applicability conditions”) that may be only partly under the control of the network node.The applicability of a UE-side AI / ML functionality might be a very dynamic property, depending, for example, on whether the UE has an AI / ML model that is applicable given the current location of the UE (e.g., geographical location, or site / cell to which the UE is connected), given the specific radio condition that the UE is experiencing, given the current speed of the UE, or given other inputs / measurements performed by device-specific sensors or algorithms.

[0014] Some of these factors may not be able to be controlled / known by the network node, because typically it is assumed that the UE-side model is not trained and generated by the network node (rather, it is typically assumed that the UE-side model is trained and generated by a node outside the RAN, such as an over-the-top (OTT) server or content network (CN) function controlled by the UE-vendor or by the mobile network operator (MNO)).

[0015] In general, the applicability of an AI / ML model / functionality to perform inference under certain conditions depends on whether the AI / ML model / functionality has been trained under such conditions. If this is the case, the performances of an AI / ML-based inference scheme can outperform conventional methods. For example, if network conditions, i.e., network (NW)-side additional conditions (such as transmitting power, antenna configuration, deployment, and / or SSB configuration) at the time in which the UE is performing the inference do not match the network conditions during the training (e.g., performed by the UE at a previous point in time), then it is likely that the AI / ML performances will be worse than the performances achieved via conventional methods. As a result, AI / ML-based beam management may lead to poor / inaccurate results.

[0016] Two types of applicability reporting were identified during the 3 GPP Rel-18 study item and are currently being discussed in RAN2 for the normative phase. These are the reactive approach and the proactive approach, as further detailed in, e.g., 3GPP TR 38.843 V18.0.0.

[0017] In the proactive approach, the network, e.g., via a network node, determines the UE capabilities and configures the UE to report the applicability of an AI / ML functionality and, based on the reported information the network configures the UE with an inference configuration, as shown in FIG. 1, which depicts the following steps:

[0018] Step 1: Network sends UECapabilityEnqiry message to initiate the procedure to a UE reporting its AI / ML supported functionalities

[0019] Step 2: UE sends UECapablity Information message to network, containing supported functionalities at the UE sideStep 3: Network configures UE that it is allowed to provide its applicable functionalities

[0020] Step 4: UE sends applicable functionalities to network upon change of applicable functi onality / condi ti on

[0021] Step 5: Network sends inference configuration for the applicable functionalities to the UE

[0022] Step 6: Start inference / monitoring based on network / UE activation / deactivation

[0023] In the reactive approach, as depicted in FIG. 2, the network, e.g., via a network node, determines the UE capabilities and configures the UE with the AI / ML functionality (possibly including the inference configuration). In response, the UE is able to determine the applicability of the A / ML functionality and, when the configured AI / ML functionality is applicable, the functionality could be active as soon as possible, without the need for additional reconfiguration. This is shown in FIG. 2:

[0024] Step 1: Network sends UECapabilityEnquiry message to initiate the procedure to a UE reporting its AI / ML supported functionalities.

[0025] Step 2: UE sends UECapablity Information message to network, containing supported functionalities at the UE side.

[0026] Step 3: Network provides network configurations and initiates UE to report its applicable functionalities.

[0027] Step 4: UE sends applicable functionalities to network.

[0028] Step 5: Network sends updated inference configuration for applicable functionalities reported in Step 4 to the UE.

[0029] Step 6: Start inference / monitoring based on network / UE activation / deactivation.

[0030] Beam Prediction

[0031] The use case of beam prediction, which may be standardized, e.g., as part of 3GPP Rel-19 work item, includes spatial beam prediction, and temporal beam prediction. The aim of this use case is to predict the “best” beam (or beams) from a Set A of beams using measurement results from another SetB of beams.

[0032] According to, e.g., 3GPP TR 38.843 vl8.0.0, the spatial -domain beam prediction for Set A of beams is based on measurement results of Set B of beams, whereas the temporal beam prediction for Set A of beams is based on the historic measurement results of SetB of beams.Hence, the radio measurements on the Set B of resources would be the input of an AI / ML model / functionality, whereas the radio measurements on the set A of resources would be the output of the AI / ML model / functionality.

[0033] Set A and Set B of beams have not been defined yet (i.e., they are left for future study); however, the following two examples illustrate some scenarios that were studied in Rel-18:

[0034] SetB is a subset of a Set A. For example, Set A is a set of eight SSB / CSL RS beams shown in FIG. 3 (both light and dark circles). The UE measures Set B (the 4 beams indicated by dark circles). The AI / ML model should predict the best beam (or beams) in Set A using only measurements from Set B. FIG. 3 depicts an example where Set B is a subset of Set A. A grid-of-beam type radiation pattern is illustrated. Each row (respective column) depicts a certain zenith (respective azimuth) angle from the antenna array. Set A has eight beams, and Set B has four beams (indicated by dark circles).

[0035] Set A and SetB correspond to two different sets of beams. In the example of FIG. 4, Set A is a set of 30 narrow CSLRS beams, and Set B is a set of eight wide SSB beams. The UE measures beams in Set B, and the AI / ML model should predict the best beam(s) from Set A.

[0036] The beam prediction can be performed in the network node and in the UE, and the gain is twofold. From the UE point of view, the UE may be able to generate good radio measurement estimations without necessarily measuring certain resources, thereby saving energy. From the network node point of view, the network node can get good radio measurements estimation from the UE without providing the measuring resources, thereby limiting the overhead over the air-interface.

[0037] Whether the UE can perform the beam prediction on a certain set of resources with a certain accuracy may depend on the applicability conditions of an AI / ML model / function. In particular, an AI / ML model / function may be trained to perform the beam prediction under certain applicability conditions. Such applicability conditions may need to be fulfilled for the AI / ML model / function to generate the expected output, (i.e., beam prediction for this use case), with enough accuracy. The applicability conditions may include a set of parameters / variables under which the AI / ML model / function was trained. Such set may include, for example, UE-specific conditions under which the model was trained, such as the UE speed, the UE antenna shape, UE sensors information such as UE orientation, and / or motion sensors. Some other parameters / variables may depend on the specific network configuration under which the model was trained, e.g., thedeployment scenario (e.g. indoor / outdoor), the carrier frequency, the network node transmission (TX) port number, the network node TX power, etc.

[0038] To determine whether an AI / ML model / function is applicable, the UE may need to assess the applicability conditions of the AI / ML model / function with respect to the output (beam prediction) that may need to be generated as well as the received input (e.g. radio measurement resources configured by the network node).

[0039] Different physical layer use cases have been studied in 3 GPP including beam management, CSI prediction, and positioning. In beam management, the UE is expected to report, via uplink control information (UCI), predictions related to the strongest beams (either ID of the strongest beams, or additionally the corresponding predicted reference signal received power (RSRP)). For CSI prediction, the UE reports predicted future CSI reports to the NW via UCI.

[0040] Once a given UE is configured with an AI / ML model / functionality, then the UE is expected to report the requested radio measurement predictions to the NW, e.g., a network node, via UCI. There are situations where the UE could fail to provide the requested reports, or the UE may want to request that the network deactivate the AL / ML model / functionality due to performance issues. 3GPP Technical Specification Group Radio Access Network (TSG RAN), RAN WG2 (RAN2) promulgated following in RAN2 128:

[0041] 1. When a functionality configured by the network to be reported via UAI, becomes from non- applicable to applicable, the UE can report it to the

[0042] network. The detailed design is for further study.

[0043] 2. When a functionality becomes non- applicable the UE doesn’t autonomously deactivate. The

[0044] NW is expected to deactivate active functionality when it

[0045] receives a report from UE that it is non-applicable.

[0046] SUMMARY

[0047] The above-described agreement in 3 GPP RAN2 128 indicates that the UE may report non-applicability of the AI / ML functionality (equivalent to the UE not being able to generate reliable predictions anymore) using RRC signalling, and then the UE may expect the NW to deactivate the functionality in response. Using RRC signaling to deactivate the AI / ML functionality may be the correct approach to inform the network that the AI / MLfunctionality is not operating correctly anymore. However, this has drawbacks that may need to be addressed. First, RRC signalling may introduce delays (for requesting resources to send the request to the NW and signaling to the UE to deactivate the feature), which in turn may degrade the performance, since during these delays the prediction (also called inference) reports received from the UE are not accurate or may not be trustworthy.

[0048] Second, it is unclear how the UE should behave after the UE indicates non-applicability in RRC but before the NW deactivates the AI / ML functionality.

[0049] Another case to consider is when the UE is temporarily unable to provide the requested prediction report. In such cases, the UE behavior should be defined so that the NW can correctly interpret the received inference results, without necessarily deactivating completely the AI / ML functionality.

[0050] Once the Network configures the UE with the inference configuration, the UE is expected to run AI / ML model / functionality and report inference reports according to the configuration. The UE may generate radio measurement prediction without measuring set A and report those measurements via uplink control channel (UCI).

[0051] According to an aspect method implemented in a UE is presented. The UE is configured to communicate with a network node. The method comprises generating, based on a ML model / functionality, a report for one or both of an inference report and a prediction report, the report comprising an indication that the one or both of the inference report and the prediction report is one or more of invalid, out of range, and inaccurate; and transmitting the report to the network node.

[0052] According to an aspect a method implemented in a network node is presented. The network node is configured to communicate with a user equipment. The method comprises configuring the UE to generate, based on a ML model / functionality, a report for one or both of an inference and a prediction, the report comprising an indication that the one or both of the inference and the prediction is one or more of invalid, out of range, and inaccurate; and receiving the report from the UE.

[0053] Described herein are example embodiments relating to transmission of an indication in the uplink control information indicating that the inference report (i.e., AI / ML prediction report) is invalid, invalid or out of range, or inaccurate.

[0054] Benefits of embodiments described herein include mitigating a UE reporting invalid or out of range AI / ML prediction results, which may lead to a performance drop since the NW will use such inaccurate information in the radio network operations (e.g.,to perform beam management with inaccurate information causing the UE to be served with a non-optimal beam). Further advantages may include:

[0055] Ensuring the accurate interpretation of inference at the NW, particularly when inference reports are invalid (e.g., low accuracy) or out of range, enabling the NW to discard such reports when necessary.

[0056] Facilitating, via a low-latency approach, rapid NW adaptation to deactivate AI / ML features when required, thereby preventing network performance degradation caused by inaccurate or invalid inference reports

[0057] This may enable quick fallback operation to a non-ML based configuration.

[0058] BRIEF DESCRIPTION OF THE DRAWINGS

[0059] A more complete understanding of the present embodiments, and the attendant advantages and features thereof, will be more readily understood by reference to the following detailed description when considered in conjunction with the accompanying drawings wherein:

[0060] FIG. l is a flowchart of an example of proactive reporting of applicability;

[0061] FIG. 2 is a flowchart of an example of reactive reporting of applicability FIG. 3 is a diagram of example SSB / CSLRS beams;

[0062] FIG. 4: is a diagram of an example set of narrow beams and set of wide beams; FIG. 5 is a schematic diagram of an example network architecture illustrating a communication system according to principles disclosed herein;

[0063] FIG. 6 is a block diagram of a network node in communication with a user equipment over a wireless connection according to some embodiments of the present disclosure;

[0064] FIG. 7 is a flowchart of an example process in a network node according to some embodiments of the present disclosure;

[0065] FIG. 8 is a flowchart of an example process in a user equipment according to some embodiments of the present disclosure;

[0066] FIG. 9 is a flowchart of an example process in a network node according to some embodiments of the present disclosure; and

[0067] FIG. 10 is a flowchart of an example process in a user equipment according to some embodiments of the present disclosure.

[0068] DETAILED DESCRIPTIONBefore describing in detail exemplary embodiments, it is noted that the embodiments reside primarily in combinations of apparatus components and processing steps related to indicating an invalid inference / prediction report. Accordingly, components have been represented where appropriate by conventional symbols in the drawings, showing only those specific details that are pertinent to understanding the embodiments so as not to obscure the disclosure with details that will be readily apparent to those of ordinary skill in the art having the benefit of the description herein.

[0069] As used herein, relational terms, such as “first” and “second,” “top” and “bottom,” and the like, may be used solely to distinguish one entity or element from another entity or element without necessarily requiring or implying any physical or logical relationship or order between such entities or elements. The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the concepts described herein. As used herein, the singular forms “a”, “an” and “the” are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will be further understood that the terms “comprises,” “comprising,” “includes” and / or “including” when used herein, specify the presence of stated features, integers, steps, operations, elements, and / or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof.

[0070] In embodiments described herein, the joining term, “in communication with” and the like, may be used to indicate electrical or data communication, which may be accomplished by physical contact, induction, electromagnetic radiation, radio signaling, infrared signaling or optical signaling, for example. One having ordinary skill in the art will appreciate that multiple components may interoperate and modifications and variations are possible of achieving the electrical and data communication.

[0071] In some embodiments described herein, the term “coupled,” “connected,” and the like, may be used herein to indicate a connection, although not necessarily directly, and may include wired and / or wireless connections.

[0072] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the concepts described herein. As used herein, the singular forms “a”, “an” and “the” are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will be further understood that the terms “comprises,” “comprising,” “includes” and / or “including” when used herein, specify the presence of stated features, integers, steps, operations, elements, and / orcomponents, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof.

[0073] The term “network node” used herein can be any kind of network node comprised in a radio network which may further comprise any of base station (BS), radio base station, base transceiver station (BTS), base station controller (BSC), radio network controller (RNC), g Node B (gNB), evolved Node B (eNB or eNodeB), Node B, multistandard radio (MSR) radio node such as MSR BS, multi -cell / multicast coordination entity (MCE), relay node, donor node controlling relay, radio access point (AP), transmission points, transmission nodes, Remote Radio Unit (RRU) Remote Radio Head (RRH), a core network node (e.g., mobile management entity (MME), self-organizing network (SON) node, a coordinating node, positioning node, MDT node, etc.), an external node (e.g., 3rd party node, a node external to the current network), nodes in distributed antenna system (DAS), a spectrum access system (SAS) node, an element management system (EMS), etc. The network node may also comprise test equipment. The term “radio node” used herein may be used to also denote a user equipment (UE) such as a wireless device (WD) or a radio network node.

[0074] In some embodiments, the non-limiting terms wireless device (WD) or a user equipment (UE) are used interchangeably. The UE herein can be any type of user equipment capable of communicating with a network node or another UE over radio signals, such as a wireless device (WD). The UE may also be a radio communication device, target device, device to device (D2D) UE, machine type UE or UE capable of machine to machine communication (M2M), low-cost and / or low-complexity UE, a sensor equipped with UE, Tablet, mobile terminals, smart phone, laptop embedded equipped (LEE), laptop mounted equipment (LME), USB dongles, Customer Premises Equipment (CPE), an Internet of Things (loT) device, or a Narrowband loT (NB-IOT) device etc.

[0075] Also, in some embodiments the generic term “radio network node” is used. It can be any kind of a radio network node which may comprise any of base station, radio base station, base transceiver station, base station controller, network controller, RNC, evolved Node B (eNB), Node B, gNB, Multi-cell / multicast Coordination Entity (MCE), relay node, access point, radio access point, Remote Radio Unit (RRU) Remote Radio Head (RRH).

[0076] Note that although terminology from one particular wireless system, such as, for example, 3GPP LTE and / or New Radio (NR), may be used in this disclosure, this should not be seen as limiting the scope of the disclosure to only the aforementioned system.Other wireless systems, including without limitation Wide Band Code Division Multiple Access (WCDMA), Worldwide Interoperability for Microwave Access (WiMax), Ultra Mobile Broadband (UMB) and Global System for Mobile Communications (GSM), may also benefit from exploiting the ideas covered within this disclosure.

[0077] According to one or more embodiments of this aspect, the general description elements in the form of “one of A and B” corresponds to A or B. According to one or more embodiments of this aspect, at least one of A and B corresponds to A, B or AB, or to one or more of A and B, or one or both of A and B . According to one or more embodiments of this aspect, at least one of A, B and C corresponds to one or more of A, B and C, and / or A, B, C or a combination thereof.

[0078] Note further, that functions described herein as being performed by a user equipment or a network node may be distributed over a plurality of user equipments and / or network nodes. In other words, it is contemplated that the functions of the network node and user equipment described herein are not limited to performance by a single physical device and, in fact, can be distributed among several physical devices.

[0079] Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this disclosure belongs. It will be further understood that terms used herein should be interpreted as having a meaning that is consistent with their meaning in the context of this specification and the relevant art and will not be interpreted in an idealized or overly formal sense unless expressly so defined herein.

[0080] Some embodiments are directed to indicating an invalid inference / prediction report.

[0081] Referring again to the drawing figures, in which like elements are referred to by like reference numerals, there is shown in FIG. 5 a schematic diagram of a communication system 10, according to an embodiment, such as a 3 GPP -type cellular network that may support standards such as LTE and / or NR (5G), which comprises an access network 12, such as a radio access network, and a core network 14. The access network 12 comprises a plurality of network nodes 16a, 16b, 16c (referred to collectively as network nodes 16), such as NBs, eNBs, gNBs or other types of wireless access points, each defining a corresponding coverage area 18a, 18b, 18c (referred to collectively as coverage areas 18). Each network node 16a, 16b, 16c is connectable to the core network 14 over a wired or wireless connection 20. A first user equipment (UE) 22a located in coverage area 18a is configured to wirelessly connect to, or be paged by, the corresponding network node 16a.A second UE 22b in coverage area 18b is wirelessly connectable to the corresponding network node 16b. While a plurality of UEs 22a, 22b (collectively referred to as user equipments 22) are illustrated in this example, the disclosed embodiments are equally applicable to a situation where a sole UE is in the coverage area or where a sole UE is connecting to the corresponding network node 16. Note that although only two UEs 22 and three network nodes 16 are shown for convenience, the communication system may include many more UEs 22 and network nodes 16.

[0082] Also, it is contemplated that a UE 22 can be in simultaneous communication and / or configured to separately communicate with more than one network node 16 and more than one type of network node 16. For example, a UE 22 can have dual connectivity with a network node 16 that supports LTE and the same or a different network node 16 that supports NR. As an example, UE 22 can be in communication with an eNB for LTE / E-UTRAN and a gNB for NR / NG-RAN.

[0083] A network node 16 (eNB or gNB) is configured to include a configuration unit 24 which is configured to perform one or more network node 16 functions described herein, including functions related to indicating an invalid inference / prediction report. A user equipment 22 is configured to include an implementation unit 26 which is configured to perform one or more UE 22 functions described herein, including functions related to indicating an invalid inference / prediction report.

[0084] Example implementations, in accordance with an embodiment, of the UE 22 and network node 16 discussed in the preceding paragraphs will now be described with reference to FIG. 6.

[0085] The communication system 10 includes a network node 16 provided in a communication system 10 and including hardware 28 enabling it to communicate with the UE 22. The hardware 28 may include a radio interface 30 for setting up and maintaining at least a wireless connection 32 with a UE 22 located in a coverage area 18 served by the network node 16. The radio interface 30 may be formed as or may include, for example, one or more RF transmitters, one or more RF receivers, and / or one or more RF transceivers. The radio interface 30 includes an array of antennas 34 to radiate and receive signal(s) carrying electromagnetic waves.

[0086] In the embodiment shown, the hardware 28 of the network node 16 further includes processing circuitry 36. The processing circuitry 36 may include a processor 38 and a memory 40. In particular, in addition to or instead of a processor, such as a central processing unit, and memory, the processing circuitry 36 may comprise integratedcircuitry for processing and / or control, e.g., one or more processors and / or processor cores and / or FPGAs (Field Programmable Gate Array) and / or ASICs (Application Specific Integrated Circuitry) adapted to execute instructions. The processor 38 may be configured to access (e.g., write to and / or read from) the memory 40, which may comprise any kind of volatile and / or nonvolatile memory, e.g., cache and / or buffer memory and / or RAM (Random Access Memory) and / or ROM (Read-Only Memory) and / or optical memory and / or EPROM (Erasable Programmable Read-Only Memory).

[0087] Thus, the network node 16 further has software 42 stored internally in, for example, memory 40, or stored in external memory (e.g., database, storage array, network storage device, etc.) accessible by the network node 16 via an external connection. The software 42 may be executable by the processing circuitry 36. The processing circuitry 36 may be configured to control any of the methods and / or processes described herein and / or to cause such methods, and / or processes to be performed, e.g., by network node 16.

[0088] Processor 38 corresponds to one or more processors 38 for performing network node 16 functions described herein. The memory 40 is configured to store data, programmatic software code and / or other information described herein. In some embodiments, the software 42 may include instructions that, when executed by the processor 38 and / or processing circuitry 36, causes the processor 38 and / or processing circuitry 36 to perform the processes described herein with respect to network node 16. For example, processing circuitry 36 of the network node 16 may include configuration unit 24 which is configured to perform one or more network node 16 functions described herein, including functions related to indicating an invalid inference / prediction report.

[0089] The communication system 10 further includes the UE 22 already referred to. The UE 22 may have hardware 44 that may include a radio interface 46 configured to set up and maintain a wireless connection 32 with a network node 16 serving a coverage area 18 in which the UE 22 is currently located. The radio interface 46 may be formed as or may include, for example, one or more RF transmitters, one or more RF receivers, and / or one or more RF transceivers. The radio interface 46 includes an array of antennas 48 to radiate and receive signal(s) carrying electromagnetic waves.

[0090] The hardware 44 of the UE 22 further includes processing circuitry 50. The processing circuitry 50 may include a processor 52 and memory 54. In particular, in addition to or instead of a processor, such as a central processing unit, and memory, the processing circuitry 50 may comprise integrated circuitry for processing and / or control, e.g., one or more processors and / or processor cores and / or FPGAs (Field ProgrammableGate Array) and / or ASICs (Application Specific Integrated Circuitry) adapted to execute instructions. The processor 52 may be configured to access (e.g., write to and / or read from) memory 54, which may comprise any kind of volatile and / or nonvolatile memory, e.g., cache and / or buffer memory and / or RAM (Random Access Memory) and / or ROM (Read-Only Memory) and / or optical memory and / or EPROM (Erasable Programmable Read-Only Memory).

[0091] Thus, the UE 22 may further comprise software 56, which is stored in, for example, memory 54 at the UE 22, or stored in external memory (e.g., database, storage array, network storage device, etc.) accessible by the UE 22. The software 56 may be executable by the processing circuitry 50. The software 56 may include a client application 58. The client application 58 may be operable to provide a service to a human or non-human user via the UE 22.

[0092] The processing circuitry 50 may be configured to control any of the methods and / or processes described herein and / or to cause such methods, and / or processes to be performed, e.g., by UE 22. The processor 52 corresponds to one or more processors 52 for performing UE 22 functions described herein. The UE 22 includes memory 54 that is configured to store data, programmatic software code and / or other information described herein. In some embodiments, the software 56 and / or the client application 58 may include instructions that, when executed by the processor 52 and / or processing circuitry 50, causes the processor 52 and / or processing circuitry 50 to perform the processes described herein with respect to UE 22. For example, the processing circuitry 50 of the user equipment 22 may include implementation unit 26 which is configured to perform one or more UE 22 functions described herein, including functions related to indicating an invalid inference / prediction report.

[0093] In some embodiments, the inner workings of the network node 16 and UE 22 may be as shown in FIG. 6 and independently, the surrounding network topology may be that of FIG. 5.

[0094] The wireless connection 32 between the UE 22 and the network node 16 is in accordance with the teachings of the embodiments described throughout this disclosure. More precisely, the teachings of some of these embodiments may improve the data rate, latency, and / or power consumption and thereby provide benefits such as reduced user waiting time, relaxed restriction on file size, better responsiveness, extended battery lifetime, etc. In some embodiments, a measurement procedure may be provided for thepurpose of monitoring data rate, latency and other factors on which the one or more embodiments improve.

[0095] Although FIGS. 5 and 6 show various “units” such as configuration unit 24 and implementation unit 26 as being within a respective processor, it is contemplated that these units may be implemented such that a portion of the unit is stored in a corresponding memory within the processing circuitry. In other words, the units may be implemented in hardware or in a combination of hardware and software within the processing circuitry.

[0096] FIG. 7 is a flowchart of an example process in a network node 16 according to some embodiments of the present disclosure. One or more blocks described herein may be performed by one or more elements of network node 16 such as by one or more of processing circuitry 36 (including the configuration unit 24), processor 38, and / or radio interface 30. Network node 16 configured to configure the UE to generate, based on an artificial intelligence / machine learning (AI / ML) model / functionality, a report for one or both of an inference (i.e. an inference report) and a prediction (i.e. a prediction report), the report comprising an indication that the one or both of the inference and the prediction is one or more of invalid, out of range, and inaccurate (Block SI 00). Network node 16 configured to receive the report from the UE (Block SI 02).

[0097] In some embodiments, the report comprises radio measurement predictions of measurement resources that comprise one or more of: prediction results for at least one measurement quantity comprising one or more of layer 1 reference signal received power, reference signal received quality (RSRQ), layer 1 signal to interference noise ratio (SINR), and received signal strength indicator (RS SI); prediction results for one or more best beam predictions; and radio measurement prediction results for one or more occasions in time.

[0098] In some embodiments, the indication indicates the one or both of the inference and the prediction is one or more of invalid, out of range, and inaccurate is based on one or more of: at least one applicability condition of the AI / ML model / functionality; performance of the AI / ML model / functionality; inability of the AI / ML model / functionality to be executed at the UE 22; an inaccuracy in the report; the report comprising a predicted quantity that is out of range; measurement inputs to the AI / ML model / functionality being fewer than a predetermined quantity; and the UE 22 requesting deactivation of the AI / ML model / functionality and not receiving an indication from the network node to deactivate the AI / ML model / functionality.In some embodiments, the indication comprises one or more of: a request from the UE 22 to deactivate the AI / ML model / functionality; and a reason the report is at least one of invalid and out of range.

[0099] In some embodiments, network node 16 is further configured to transmit another indication to the UE 22 to deactivate the AI / ML model / functionality.

[0100] FIG. 8 is a flowchart of an example process in a user equipment 22 according to some embodiments of the present disclosure. One or more blocks described herein may be performed by one or more elements of user equipment 22 such as by one or more of processing circuitry 50 (including the implementation unit 26), processor 52, and / or radio interface 46. UE 22 is configured to generate, based on an AI / ML model / functionality, a report for one or both of an inference and a prediction, the report comprising an indication that the one or both of the inference and the prediction is one or more of invalid, out of range, and inaccurate (Block SI 04). UE 22 is configured to transmit the report to the network node 16 (Block SI 06).

[0101] In some embodiments, the report comprises radio measurement predictions of measurement resources that comprise one or more of: prediction results for at least one measurement quantity comprising one or more of layer 1 reference signal received power, RSRQ, layer 1 SINR, and RSSI; prediction results for one or more best beam predictions; and radio measurement prediction results for one or more occasions in time.

[0102] In some embodiments, the indication indicates the one or both of the inference and the prediction is one or more of invalid, out of range, and inaccurate is based on one or more of: at least one applicability condition of the AI / ML model / functionality; performance of the AI / ML model / functionality; inability of the AI / ML model / functionality to be executed at the UE 22; an inaccuracy in the report; the report comprising a predicted quantity that is out of range; measurement inputs to the AI / ML model / functionality being fewer than a predetermined quantity; and the UE 22 requesting deactivation of the AI / ML model / functionality and not receiving an indication from the network node to deactivate the AI / ML model / functionality.

[0103] In some embodiments, the indication comprises one or more of: a request from the UE 22 to deactivate the AI / ML model / functionality; and a reason the report is at least one of invalid and out of range.

[0104] In some embodiments, UE 22 is configured to receive another indication from the network node 16 to deactivate the AI / ML model / functionality.FIG. 9 is a flowchart of an example process in a network node 16 according to some embodiments of the present disclosure. One or more blocks described herein may be performed by one or more elements of network node 16 such as by one or more of processing circuitry 36 (including the configuration unit 24), processor 38, and / or radio interface 30. The process is implemented in a network node that is configured to communicate with a UE. The process comprising configuring (S200) the UE to generate, based on a ML model / functionality, a report for one or both of an inference and a prediction, the report comprising an indication that the one or both of the inference and the prediction is one or more of invalid, out of range, and inaccurate; and receiving (S202) the report from the UE.

[0105] In some embodiments the report comprises radio measurement predictions of measurement resources that comprise one or more of: prediction results for at least one measurement quantity comprising one or more of layer 1 reference signal received power, reference signal received quality (RSRQ), layer 1 signal to interference noise ratio (SINR), and received signal strength indicator (RS SI); prediction results for one or more best beam predictions; and radio measurement prediction results for one or more occasions in time.

[0106] In some embodiments the indication indicates the one or both of the inference and the prediction is one or more of invalid, out of range, and inaccurate is based on one or more of: at least one applicability condition of the ML model / functionality; performance of the ML model / functionality; inability of the ML model / functionality to be executed at the UE; an inaccuracy in the report; the report comprising a predicted quantity that is out of range; measurement inputs to the ML model / functionality being fewer than a predetermined quantity; and the UE requesting deactivation of the ML model / functionality and not receiving an indication from the network node to deactivate the ML model / functionality.

[0107] In some embodiments the indication comprises one or more of: a request from the UE to deactivate the ML model / functionality; and a reason the report is at least one of invalid and out of range.

[0108] In some embodiments the method comprises transmitting another indication to the UE to deactivate the ML model / functionality.

[0109] In some embodiments the process further comprises: receiving (S203) an indication from the UE, indicating a reason the report is one or more of invalid, out of range, and inaccurate. The indication may be received from the UE after the report is received.FIG. 10 is a flowchart of an example process in a user equipment 22 according to some embodiments of the present disclosure. One or more blocks described herein may be performed by one or more elements of user equipment 22 such as by one or more of processing circuitry 50 (including the implementation unit 26), processor 52, and / or radio interface 46. The process is implemented in a UE. The UE is configured to communicate with a network node. The method comprises: generating (S204), based on a ML model / functionality, a report for one or both of an inference report and a prediction report, the report comprising an indication that the one or both of the inference report and the prediction report is one or more of invalid, out of range, and inaccurate; and transmitting (S206) the report to the network node.

[0110] In some embodiments the report comprises radio measurement predictions of measurement resources that comprise one or more of: prediction results for at least one measurement quantity comprising one or more of layer 1 reference signal received power, reference signal received quality (RSRQ), layer 1 signal to interference noise ratio (SINR), and received signal strength indicator (RS SI); prediction results for one or more best beam predictions; and radio measurement prediction results for one or more occasions in time.

[0111] In some embodiments the indication indicates the one or both of the inference and the prediction is one or more of invalid, out of range, and inaccurate is based on one or more of: at least one applicability condition of the ML model / functionality; performance of the ML model / functionality; inability of the ML model / functionality to be executed at the UE; an inaccuracy in the report; the report comprising a predicted quantity that is out of range; measurement inputs to the ML model / functionality being fewer than a predetermined quantity; and the UE requesting deactivation of the ML model / functionality and not receiving an indication from the network node to deactivate the ML model / functionality.

[0112] In some embodiments the indication comprises one or more of: a request from the UE to deactivate the ML model / functionality; and a reason the report is at least one of invalid and out of range.

[0113] In some embodiments the process further comprises: transmitting (S208) an indication to the network node, indicating a reason the report is one or more of invalid, out of range, and inaccurate.

[0114] In some embodiments transmitting the indication is performed after transmitting the report.Having described the general process flow of arrangements of the disclosure and having provided examples of hardware and software arrangements for implementing the processes and functions of the disclosure, the sections below provide details and examples of arrangements for indicating an invalid inference / prediction report. One or more UE 22 functions described below may be performed by one or more of processing circuitry 50, processor 52, implementation unit 26, etc. One or more network node 16 functions described below may be performed by one or more of processing circuitry 36, processor 38, configuration unit 24, etc.

[0115] In some embodiments, “higher layer signalling” may refer to RRC control signalling. “NW” may refer to a RAN network node, such as the network node 16.

[0116] Step 0: configuration and activation of the AI / ML functionality

[0117] For AI / ML BM LCM operations, it has been agreed to reuse the existing procedures and terminologies from the CSI Framework, including those defined for aperiodic, semipersi stent on physical uplink control channel (PUCCH), semipersistent on physical uplink shared channel (PUSCH), and periodic reporting configuration in, e.g., RAN2#127bis. In Step 3 (discussed below) the network (e.g., via network node 16) may configure one or more of: CSI-ReportConfig for inference configuration including associated ID, Set A related information, Set B related information, inference report content related information, and / or time instances related information for measurements and prediction.

[0118] Whether the UE 22 can perform the AI / ML inference, given a certain Set A / B of resources, depends on whether the UE 22 has an available AI / ML model / functionality that can perform the inference / prediction on those Set A of resources, given the measurements on the Set B of resources. In other words, the UE 22 can perform the AI / ML inference depending on whether the AI / ML model / functionality is applicable under that Set A / B configuration. Once the UE 22 confirms the applicability of the AI / ML model / functionality, it acknowledges the activation of the inference configuration in a RRCReconfigurationComplete message. For aperiodic and semi-persistent, the activation is via L1 / L2. The UE 22 may then be expected to provide inference / prediction reports according to the information indicated in the inference configuration. For instance, the inference / prediction reports could be sent semi-persistently, periodically, or a-periodically in response to a trigger in downlink control information (DCI). The report may be carried on UCI in either PUSCH or PUCCH.

[0119] Step 1: inability to provide the requested inference reportDetails on the inability to provide inference report results

[0120] The UE 22 may be unable to provide the inference reports for various reasons, including the following non-limiting examples:

[0121] • AI / ML model / functionality is no longer applicable, e.g,. the training conditions do not match the conditions in which inference is being executed.

[0122] • Low output accuracy / high output uncertainty provided by the AI / ML model / functionality and, based on such evaluation, deciding that the performance does not fulfill the requirements.

[0123] o The low output accuracy can, for example, be determined using monitoring resources configured at the UE 22 by the NW (e.g., via network node 16). Based on such monitoring configuration, the UE 22 can determine whether the model output accuracy is low.

[0124] • The AI / ML model / functionality available at the UE 22 cannot be executed / run (e.g., battery, computation or storage constraints)

[0125] • The computed inference report is inaccurate (i.e., of low confidence) o For example, the confidence in the output is below a certain threshold level, for example in beam management use case, the highest probability of any of the predicted beams being the Top-1 beam is below a certain threshold. The threshold can be configured by the NW (e.g., via network node 16) in some embodiments, or may be pre-defined in standards.

[0126] • Lack of inputs to the model, such as measurements of reference signal(s) o For example, the UE 22 may only measure a subset of the beams in set B for beam predictions. Alternatively, the UE 22 may not store all needed measurements for temporal beam prediction, hence the UE 22 may have needed to drop some measurement(s) from its memory.

[0127] o In some embodiments, UE 22 may not have a chance to measure all the beams in Set B for beam prediction due to a recent CSI report (re)configuration, serving cell activation, bandwidth part (BWP) change, or activation of semipersistent-CSI.

[0128] o Inputs other than reference signal measurements could include UE 22 location, UE 22 sensors such as inertial measurement unit (IMU), UE 22 velocity estimations, UE 22 orientation, etc.• Transmitting a request to deactivate the AI / ML model / functionality and not receiving an indication from the NW (e.g., via network node 16) to deactivate the functionality, where the deactivation request is sent via higher layer signaling.

[0129] If UE 22 is unable to provide an inference / prediction report (e.g. comprising predicted beams), the UE 22 reports a reserved value(s) in the inference / prediction report to indicate to the NW (e.g., via network node 16) the invalidity of the reported measurements.

[0130] In some cases, one or more predicted quantities (e.g., predicted layer 1 RSRP) may be out of range of the quantized values that are allowed to be reported for the respective predicted quantity. For example, considering predicted layer 1 RSRP to be the predicted quantity to be reported as part of inference / prediction report, let [0O

[0131]

[0132] ^2 ■■■ ^M-2] be the quantized values allowed to be reported for predicted layer 1 RSRP. The predicted layer 1 RSRP for one or more predicted beams may be either below the lowest allowed quantized value 0Oby a first margin or above the highest allowed quantized value 0M-2by a second margin. The UE 22 reports out of range value(s) for the predicted layer 1 RSRP for the one or more prediction beams as part of the inference / prediction report. The first and the second margins may be configured by the network to the UE 22 in some embodiments. In some embodiments, the first and the second margins may be prespecified, e.g., in 3GPP specifications. This indication of out of range value(s) for predicted layer 1 RSRP for one or more beams indicates to the network (e.g., via network node 16) that the predicted layer 1 RSRP is inaccurate (either too low or too high). In some embodiments, each predicted layer 1 RSRP is reported via a M-bit field (where M > l is a positive integer), and one of the codepoints in the M-bit field is reserved for indicating out of range value. The remaining 2M— 1 codepoints in the M-bit field are used to indicate the other allowed quantized values (i.e., 0O#i ^2 ■■■

[0133]

[0134] Example Embodiments related to the inference / prediction report format and content

[0135] In some embodiments, the UE 22 reports a reserved value(s) in the inference / prediction report to indicate to the NW (e.g., via network node 16) the invalidity or out of range nature of the reported measurements. Consider the case of UE 22 reporting predicted strongest beam identities (IDs) and corresponding predicted layer 1 RSRP as part of inference / prediction report. In some embodiments, a codepoint in a field indicating predicted strongest beam ID may be reserved to indicate that the corresponding predicted beam is invalid or out of range. In some embodiments, a codepoint in a field forindicating predicted layer 1 RSRP may be reserved to indicate that the corresponding predicted layer 1 RSRP is invalid or out of range. In some embodiments, a certain reserved value (e.g., “invalid,” “out of range,” or “NAN”) may be indicated in the reserved codepoint to indicate failure of providing the related field in the inference / prediction report. The certain reserved value(s) could be preconfigured by the NW (e.g., via network node 16) or defined / fixed in a specification, e.g., as specified by 3GPP.

[0136] If the UE 22 reports a predicted beam ID, where the ID is a CSI-RS resource indicator (CRI) in a set ResourceSet, the ID may comprise a number between [0, ..., SetAResources-1], In at least one embodiment, to also include the reserved value(s), the UE can report a value between [0, ... , SetAResources - 1 , one or more reserved Values], In some embodiments, a single reserved value is used to indicate a possible inapplicability, invalidity, or out of range condition — hence the UE 22 may report the value of #setAresouces (last value of the range of valid values) to indicate an inapplicability invalidity, or out of range condition. This could be achieved via the following extensions to UCI, via the following extensions to, e.g., 3 GPP TR 38.212:

[0137] The bitwidth for CRI, SSBRI, RSRP, differential RSRP, and Capability Index are provided in Table 6.3.1.1.2-6.

[0138] Table 6.3.1.1.2-6: CRI, SSBRI, RSRP, and Capabilityindex

[0139]

[0140] where KBSI~RSis the number of CSI-RS resources in the corresponding resource set, and KgSBis the configured number of SS / PBCH blocks in the corresponding resource set for reporting 'ssb-Index-SINR'. The extra bit in CRI and / or SSBRI is to enable the UE to report an inapplicable inference report.

[0141] In some embodiments, the UE 22 can report inapplicability in dedicated bit(s) in the UCI, via the following extensions:

[0142] The bitwidth for CRI, SSBRI, RSRP, differential RSRP, and Capability Index are provided in Table 6.3.1.1.2-6.Table 6.3.1.1.2-6: CRI, SSBRI, RSRP, and Capabilityindex

[0143]

[0144] where K^SI~RSis the number of CSI-RS resources in the corresponding resource set, and KgSBis the configured number of SS / PBCH blocks in the corresponding resource set for reporting 'ssb-Index-SINR'. The applicability bit indicates if the UE could accurately provide the other information in the inference report.

[0145] In at least one embodiment, when the top-K beams are reported in a bitmap format, the UE 22 could report all / most beams as “1” to indicate the inapplicability. In this case, the NW (e.g., via network node 16) therefore receives an indication that all / most beams are part of top-K, indicating that the prediction is unreliable.

[0146] In some embodiments, the UE 22 might only be capable of predicting the top-K beam IDs, and not the RSRP of the corresponding beams. In this case, the UE 22 could report the minimum RSRP value, or the UCI could be extended with an “inapplicable” RSRP report. Hence the bitwidth of RSRP could be eight instead of seven in the 3 GPP Table 6.3.1.1.2-6 reproduced above.

[0147] In some embodiments, the UE 22 can indicate invalidity or out of range condition for part of the inference / prediction report is for the temporal beam prediction. The UE 22 might be configured to report predicted best beam information for multiple future time instances. The UE 22 indicates that, for certain future time instances, the reported information is invalid or out of range because for example, a certain model output accuracy could not be guaranteed.

[0148] In some embodiments, the UE 22 may report a low accuracy value to indicate that the measurement predictions associated to the predicted beam is invalid, out of range, or inaccurate. For example, if the accuracy of a certain prediction is below a configured threshold, then the UE 22 may include a low accuracy value associated to thecorresponding prediction. The low accuracy value can be included within the UCI in a field used by the UE 22 to report the accuracy of the prediction.

[0149] In some embodiments, it could be multiple values reserved wherein each reserved value indicates a different cause of failure to provide the requested inference / prediction report.

[0150] Step 2: deactivation or reconfiguration of the AI / ML model / functionality

[0151] The NW (e.g., via network node 16) receives the indication of invalidity of all or part of the requested inference / prediction report.

[0152] In at least one embodiment, the NW (e.g., via network node 16) may discard the invalid information in the inference / prediction report. The NW (e.g., via network node 16) may continue to discard the inference / prediction reports until the UE 22 indicates that the inference information is valid again.

[0153] In some embodiments, the NW (e.g., via network node 16) may reconfigure / request the UE 22 to provide the requested information based on actual radio measurements instead of AI / ML inference, until the inference information is valid again.

[0154] In some embodiments, the NW (e.g., via network node 16) may also perform a reconfiguration, e.g., change from a UE 22 report of Top-K beam IDs and RSRP to only report beam IDs, for the UE 22 to report for another future time instance, or for the UE 22 to report for another resource set A.

[0155] Alternatively, the NW (e.g., via network node 16) receives the indication and an explicit request from the UE 22 to deactivate the AI / ML functionality due to change in the applicability status, i.e., that the AI / ML functionality is no longer applicable.

[0156] In response, the UE receives from the NW (e.g., via network node 16) a request to deactivate or de-configure the relevant AI / ML functionality, or the concerned inference reporting configuration associated to an AI / ML functionality. In one embodiment, the NW determines to deactivate or deconfigure the said AI / ML functionality or reporting configuration, upon receiving a certain number of UCI inference reports including invalid, or out-of-range, or inaccuracy indications, e.g. within a time window, or a number of successive UCI inference reports including invalid, or out-of-range, or inaccuracy indications.

[0157] Information in inference / prediction report, a step that may take place prior to or after Step 2In at least one embodiment, the UE 22 further reports information indicating why an inference / prediction report could not be provided. The UE 22 could report such information:

[0158] As part of the inference / prediction report

[0159] In a separate message after the inference / prediction report. This could enable the UE 22 to aggregate information regarding the reasons why one or more inference / prediction reports could not be provided. This message could, for example, be indicated by the UE 22 in a UE 22 Assistance Information (UAI) message. The NW (e.g., via network node 16) could configure the UE 22 to provide such information in the RRC (re)configuration of the inference / prediction report configuration.

[0160] Possible reasons why the UE 22 may have been unable to provide the inference / prediction report include:

[0161] • AEML model / functionality is no longer applicable. For example, the training conditions do not match the conditions in which inference is being executed.

[0162] • Low output accuracy provided by the of the AI / ML model / functionality (e.g., determined from a performance monitoring operation at the UE 22)

[0163] • The AI / ML model / functionality available at the UE 22 cannot be executed / run (e.g., battery, computation or storage constraints)

[0164] • The computed inference / prediction report is inaccurate (of low confidence)

[0165] • Lack of required inputs to the model, e.g., UE 22 measurements.

[0166] Rules for UE to determine whether / when to send also higher layer signaling in addition to Laver 1 UCI signaling

[0167] To avoid sending duplicate information about an AI / ML functionality being non-applicable from the UE 22 to the NW (e.g., via network node 16), the UE 22 could be provided with some rules to determine whether and when (i.e., the triggers) to send higher layer signaling (in UAI in RRC), in addition to the indication in UCI about the inference / prediction report being invalid.

[0168] The rules for the UE 22 can be written in specification text (implicitly the same rules for all UEs 22), or can be configured by the NW (e.g., via network node 16) for each UE 22.

[0169] Examples of such rules are:

[0170] • The UE 22 may send higher layer signaling only if it represents / contains an explicit request to the NW (e.g., via network node 16) to deactivate the AI / MLfunctionality, upon which the NW (e.g., via network node 16) should immediately deactivate that AI / ML functionality. The request may indicate the AI / ML functionalities that are no longer applicable to the UE 22, or may indicate an identity of the inference reporting configuration that are no longer applicable at the UE 22. If higher layer signaling is not sent by the UE 22, and only the UCI indicates an invalid inference / prediction report, the NW (e.g., via network node 16) implicitly assumes that the UE 22 can only temporarily not provide the inference / prediction report and is to expect the inference / prediction reports to be valid again soon. Hence, the NW (e.g., via network node 16) may decide not to deactivate the AI / ML functionality. This rule is useful especially if the UE 22 is unable to explicitly indicate the reason why an inference / prediction report in UCI is invalid, and the rule helps the NW (e.g., via network node 16) to decide whether it should immediately deactivate an AI / ML functionality or whether it can wait for the UE 22 to recover.

[0171] o Alternatively, the higher layer signaling could be triggered to indicate the UE 22 preference to not deactivate an AI / ML functionality, despite invalid inference / prediction reports in UCI. This would explicitly indicate to the NW (e.g., via network node 16) that the UE 22 expects the inference / prediction reports to be valid again soon.

[0172] o For example, some embodiments, the UE 22 transmits the higher layer signalling to indicate inapplicability of the AI / ML functionality or of a reporting configuration. This may occur upon indicating in a certain number of successive UCI the invalid inference / prediction reports or upon indicating, in a certain number of UCI within a certain time window, the invalid inference / prediction reports. Such number of UCI and the time window can be configured by the network node 16.

[0173] • The UE 22 may send higher layer signaling only if the signalling contains additional information compared to the UCI with an invalid inference / prediction report, for example:

[0174] o the reason why one or more inference / prediction reports could not be provided, if this is not possible to indicate in UCI; and / or

[0175] o a UE 22 preference for another AI / ML configuration or a preference for one or more parameters in an AI / ML configuration, where such preference would make the inference / prediction reports valid again.

[0176] Some further example embodiments are provided below.Example 1. A method at a UE 22 for transmitting inference or prediction reports via uplink control information, based on an AI / ML model / functionality that is configured by the NW (e.g., via network node 16) to the UE 22, wherein the inference or prediction report comprises an indication indicating that the reported inference or prediction report is at least one of invalid or out of range, or inaccurate.

[0177] Example 2. A method according to Example 1, wherein the inference or prediction report of the AI / ML model / functionality available at the UE 22 is the radio measurement predictions of measurement resources comprising one or more of the following:

[0178] • the prediction results for one or more measurement quantities, such as the layer 1 RSRP, reference signal received quality (RSRQ), layer 1 signal to interference noise ratio (SINR), received signal strength indicator (RSSI);

[0179] • prediction results for one or more best beam predictions; and • radio measurement prediction results for one or more occasions in time.

[0180] Example 3. A method according to any of Examples 1-2, wherein the indication indicating that the reported inference or prediction report is at least one of invalid and out of range is transmitted in response to one or more of:

[0181] • evaluating the applicability conditions of the AI / ML model / functionality available at the UE 22 and based on such evaluation deciding that the AI / ML model / functionality is inapplicable;

[0182] • evaluating the performance of the AI / ML model / functionality and based on such evaluation deciding that the performance does not fulfill the requirements;

[0183] • The AI / ML model / functionality available at the UE 22 cannot be executed / run;

[0184] • The computed inference or prediction report is inaccurate;

[0185] • The one or more predicted quantities (e.g., predicted layer 1 RSRP) is out of range of the quantized values that are allowed to be reported for the one or more measurement quantities;

[0186] • Lack of measurement inputs to the model; and

[0187] • Transmitting a request to deactivate the AI / ML model / functionality and not receiving an indication from the NW (e.g., via network node 16) to deactivate the functionality, where the deactivation request is sent via higher layer signaling.Example 4. A method according to any of Examples 1-3, wherein the indication indicating that the reported inference or prediction report is at least one of invalid or out of range or inaccurate is an indication from the UE 22 to request deactivation of the AI / ML model / functionality.

[0188] Example 5. A method according to any of Examples 1-4, further comprising the value(s) reported in the inference or prediction report corresponding to one or more reserved value; and determining that the inference or prediction report is invalid or out of range, or inaccurate in accordance with the reserved value.

[0189] Example 6. A method according to Example 5, wherein each reserved value indicates information related to the reason of the inference or prediction report being at least one of invalid and out of range.

[0190] Example 7. The method according to Example 3, wherein the applicability conditions of the AI / ML model / functionality available at the UE 22 are checked by the UE 22 prior to transmitting the indication indicating that the reported inference or prediction report is at least one of invalid and out of range.

[0191] Example 8. The method according to Example 5, wherein in response of transmitting the indication, the UE 22 receives a second indication from the NW (e.g., via network node 16) indicating that AI / ML model / functionality should be deactivated, and UE should not provide inference or prediction reports for the deactivated AI / ML model / functionality.

[0192] Example 9. A method according to Example 1, wherein the UE 22 is provided / configured with rules to determine whether and in which conditions the UE 22 sends also RRC signaling to indicate that the AI / ML functionality or a reporting configuration associated to a certain AI / ML functionality is non-applicable, in addition to the Layer 1 signaling in UCI indicating that the reported inference or prediction report is at least one of invalid and out of range.

[0193] Example 10: The method of Example 1, wherein in response of transmitting one or more inference or prediction reports comprising an indication indicating that the reported inference or prediction report is at least one of invalid or out of range, or inaccurate, the UE 22 receives indicating from the network node 16 to deactivate the concerned AI / ML functionality, or the concerned inference reporting configuration associated to an AI / ML functionality.

[0194] Some additional examples:Example Al . A method implemented in a user equipment (UE) that is configured to communicate with a network node, the method comprising:

[0195] generating, based on an artificial intelligence / machine learning (AI / ML) model / functionality, a report for one or both of an inference and a prediction, the report comprising an indication that the one or both of the inference and the prediction is one or more of invalid, out of range, and inaccurate; and

[0196] transmitting the report to the network node.

[0197] Example A2. The method of Example Al, wherein the report comprises radio measurement predictions of measurement resources that comprise one or more of prediction results for at least one measurement quantity comprising one or more of layer 1 reference signal received power, reference signal received quality (RSRQ), layer 1 signal to interference noise ratio (SINR), and received signal strength indicator (RSSI); prediction results for one or more best beam predictions; and

[0198] radio measurement prediction results for one or more occasions in time.

[0199] Example A3. The method of Example Al, wherein the indication indicates the one or both of the inference and the prediction is one or more of invalid, out of range, and inaccurate is based on one or more of

[0200] at least one applicability condition of the AI / ML model / functionality; performance of the AI / ML model / functionality;

[0201] inability of the AI / ML model / functionality to be executed at the UE;

[0202] an inaccuracy in the report;

[0203] the report comprising a predicted quantity that is out of range;

[0204] measurement inputs to the AI / ML model / functionality being fewer than a predetermined quantity; and

[0205] the UE requesting deactivation of the AI / ML model / functionality and not receiving an indication from the network node to deactivate the AI / ML model / functionality.

[0206] Example A4. The method of Example Al, wherein the indication comprises one or more of

[0207] a request from the UE to deactivate the AI / ML model / functionality; and a reason the report is at least one of invalid and out of range.

[0208] Example A5. The method of Example Al, wherein the method comprises receiving another indication from the network node to deactivate the AI / ML

[0209] model / functionality.Example Bl. A user equipment (UE) configured to communicate with a network node, the UE configured to, and / or comprising a radio interface and / or processing circuitry configured to:

[0210] generate, based on an artificial intelligence / machine learning (AI / ML) model / functionality, a report for one or both of an inference and a prediction, the report comprising an indication that the one or both of the inference and the prediction is one or more of invalid, out of range, and inaccurate; and

[0211] transmit the report to the network node.

[0212] Example B2. The UE of Example Bl, wherein the report comprises radio measurement predictions of measurement resources that comprise one or more of prediction results for at least one measurement quantity comprising one or more of layer 1 reference signal received power, reference signal received quality (RSRQ), layer 1 signal to interference noise ratio (SINR), and received signal strength indicator (RSSI); prediction results for one or more best beam predictions; and

[0213] radio measurement prediction results for one or more occasions in time.

[0214] Example B3. The UE of Example Bl, wherein the indication indicates the one or both of the inference and the prediction is one or more of invalid, out of range, and inaccurate is based on one or more of

[0215] at least one applicability condition of the AI / ML model / functionality; performance of the AI / ML model / functionality;

[0216] inability of the AI / ML model / functionality to be executed at the UE;

[0217] an inaccuracy in the report;

[0218] the report comprising a predicted quantity that is out of range;

[0219] measurement inputs to the AI / ML model / functionality being fewer than a predetermined quantity; and

[0220] the UE requesting deactivation of the AI / ML model / functionality and not receiving an indication from the network node to deactivate the AI / ML model / functionality.

[0221] Example B4. The UE of Example Bl, wherein the indication comprises one or more of

[0222] a request from the UE to deactivate the AI / ML model / functionality; and a reason the report is at least one of invalid and out of range.

[0223] Example B5. The UE of Example Bl, further configured to receive another indication from the network node to deactivate the AI / ML model / functionality.Example Cl . A method implemented in a network node that is configured to communicate with a user equipment, the method comprising:

[0224] configuring the UE to generate, based on an artificial intelligence / machine learning (AI / ML) model / functionality, a report for one or both of an inference and a prediction, the report comprising an indication that the one or both of the inference and the prediction is one or more of invalid, out of range, and inaccurate; and

[0225] receiving the report from the UE.

[0226] Example C2. The method of Example Cl, wherein the report comprises radio measurement predictions of measurement resources that comprise one or more of prediction results for at least one measurement quantity comprising one or more of layer 1 reference signal received power, reference signal received quality (RSRQ), layer 1 signal to interference noise ratio (SINR), and received signal strength indicator (RSSI); prediction results for one or more best beam predictions; and

[0227] radio measurement prediction results for one or more occasions in time.

[0228] Example C3. The method of Example Cl, wherein the indication indicates the one or both of the inference and the prediction is one or more of invalid, out of range, and inaccurate is based on one or more of

[0229] at least one applicability condition of the AI / ML model / functionality; performance of the AI / ML model / functionality;

[0230] inability of the AI / ML model / functionality to be executed at the UE;

[0231] an inaccuracy in the report;

[0232] the report comprising a predicted quantity that is out of range;

[0233] measurement inputs to the AI / ML model / functionality being fewer than a predetermined quantity; and

[0234] the UE requesting deactivation of the AI / ML model / functionality and not receiving an indication from the network node to deactivate the AI / ML model / functionality.

[0235] Example C4. The method of Example Cl, wherein the indication comprises one or more of

[0236] a request from the UE to deactivate the AI / ML model / functionality; and

[0237] a reason the report is at least one of invalid and out of range.

[0238] Example C5. The method of Example Cl, wherein the method comprises transmitting another indication to the UE to deactivate the AI / ML model / functionality.Example DI . A network node configured to communicate with a user equipment (UE), the network node configured to, and / or comprising a radio interface and / or comprising processing circuitry configured to:

[0239] configure the UE to generate, based on an artificial intelligence / machine learning (AI / ML) model / functionality, a report for one or both of an inference and a prediction, the report comprising an indication that the one or both of the inference and the prediction is one or more of invalid, out of range, and inaccurate; and

[0240] receive the report from the UE.

[0241] Example D2. The network node of Example DI, wherein the report comprises radio measurement predictions of measurement resources that comprise one or more of prediction results for at least one measurement quantity comprising one or more of layer 1 reference signal received power, reference signal received quality (RSRQ), layer 1 signal to interference noise ratio (SINR), and received signal strength indicator (RSSI); prediction results for one or more best beam predictions; and

[0242] radio measurement prediction results for one or more occasions in time

[0243] Example D3. The network node of Example DI, wherein the indication indicates the one or both of the inference and the prediction is one or more of invalid, out of range, and inaccurate is based on one or more of

[0244] at least one applicability condition of the AI / ML model / functionality; performance of the AI / ML model / functionality;

[0245] inability of the AI / ML model / functionality to be executed at the UE;

[0246] an inaccuracy in the report;

[0247] the report comprising a predicted quantity that is out of range;

[0248] measurement inputs to the AI / ML model / functionality being fewer than a predetermined quantity; and

[0249] the UE requesting deactivation of the AI / ML model / functionality and not receiving an indication from the network node to deactivate the AI / ML model / functionality.

[0250] Example D4. The network node of Example DI, wherein the indication comprises one or more of

[0251] a request from the UE to deactivate the AI / ML model / functionality; and a reason the report is at least one of invalid and out of range.

[0252] Example D5. The network node of Example DI, further configured to transmit another indication to the UE to deactivate the AI / ML model / functionality. Example ALA method implemented in a user equipment (UE) that is configured to communicate with a network node, the method comprising:

[0253] generating, based on an artificial intelligence / machine learning (AI / ML) model / functionality, a report for one or both of an inference and a prediction, the report comprising an indication that the one or both of the inference and the prediction is one or more of invalid, out of range, and inaccurate; and

[0254] transmitting the report to the network node.

[0255] As will be appreciated by one of skill in the art, the concepts described herein may be embodied as a method, data processing system, computer program product and / or computer storage media storing an executable computer program. Accordingly, the concepts described herein may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects all generally referred to herein as a “circuit” or “module.” Any process, step, action and / or functionality described herein may be performed by, and / or associated to, a corresponding module, which may be implemented in software and / or firmware and / or hardware. Furthermore, the disclosure may take the form of a computer program product on a tangible computer usable storage medium having computer program code embodied in the medium that can be executed by a computer. Any suitable tangible computer readable medium may be utilized including hard disks, CD-ROMs, electronic storage devices, optical storage devices, or magnetic storage devices.

[0256] Some embodiments are described herein with reference to flowchart illustrations and / or block diagrams of methods, systems and computer program products. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer (to thereby create a special purpose computer), special purpose computer, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions / acts specified in the flowchart and / or block diagram block or blocks.

[0257] These computer program instructions may also be stored in a computer readable memory or storage medium that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored inthe computer readable memory produce an article of manufacture including instruction means which implement the function / act specified in the flowchart and / or block diagram block or blocks.

[0258] The computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions / acts specified in the flowchart and / or block diagram block or blocks.

[0259] It is to be understood that the functions / acts noted in the blocks may occur out of the order noted in the operational illustrations. For example, two blocks shown in succession may in fact be executed substantially concurrently or the blocks may sometimes be executed in the reverse order, depending upon the functionality / acts involved. Although some of the diagrams include arrows on communication paths to show a primary direction of communication, it is to be understood that communication may occur in the opposite direction to the depicted arrows.

[0260] Computer program code for carrying out operations of the concepts described herein may be written in an object oriented programming language such as Python, Java® or C++. However, the computer program code for carrying out operations of the disclosure may also be written in conventional procedural programming languages, such as the "C" programming language. The program code may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer. In the latter scenario, the remote computer may be connected to the user's computer through a local area network (LAN) or a wide area network (WAN), or the connection may be made to an external computer (for example, through the Internet using an Internet Service Provider).

[0261] Many different embodiments have been disclosed herein, in connection with the above description and the drawings. It will be understood that it would be unduly repetitious and obfuscating to literally describe and illustrate every combination and subcombination of these embodiments. Accordingly, all embodiments can be combined in any way and / or combination, and the present specification, including the drawings, shall be construed to constitute a complete written description of all combinations and subcombinations of the embodiments described herein, and of the manner and process ofmaking and using them, and shall support claims to any such combination or subcombination.

[0262] It will be appreciated by persons skilled in the art that the embodiments described herein are not limited to what has been particularly shown and described herein above. In addition, unless mention was made above to the contrary, it should be noted that all of the accompanying drawings are not to scale. A variety of modifications and variations are possible in light of the above teachings.

Claims

CLAIMS1. A method implemented in a user equipment, UE, that is configured to communicate with a network node, the method comprising:generating (S204), based on a machine learning, ML, model / functionality, a report for one or both of an inference report and a prediction report, the report comprising an indication that the one or both of the inference report and the prediction report is one or more of invalid, out of range, and inaccurate; andtransmitting (S206) the report to the network node.

2. The method of Claim 1, wherein the report comprises radio measurement predictions of measurement resources that comprise one or more ofprediction results for at least one measurement quantity comprising one or more of layer 1 reference signal received power, reference signal received quality, RSRQ, layer 1 signal to interference noise ratio, SINR, and received signal strength indicator, RSSI; prediction results for one or more best beam predictions; andradio measurement prediction results for one or more occasions in time.

3. The method of Claims 1 or 2, wherein the indication indicates the one or both of the inference and the prediction is one or more of invalid, out of range, and inaccurate is based on one or more ofat least one applicability condition of the ML model / functionality; performance of the ML model / functionality;inability of the ML model / functionality to be executed at the UE;an inaccuracy in the report;the report comprising a predicted quantity that is out of range;measurement inputs to the ML model / functionality being fewer than a predetermined quantity; andthe UE requesting deactivation of the ML model / functionality and not receiving an indication from the network node to deactivate the ML model / functionality.

4. The method of any one of Claims 1-3, wherein the indication comprises one or more ofa request from the UE to deactivate the ML model / functionality; anda reason the report is at least one of invalid and out of range.

5. The method of any one of Claims 1-4, further comprising: transmitting (S208) an indication to the network node, indicating a reason the report is one or more of invalid, out of range, and inaccurate.

6. The method of Claim 5, wherein transmitting the indication is performed after transmitting the report.

7. A method implemented in a network node that is configured to communicate with a user equipment, the method comprising:configuring (S200) the UE to generate, based on a machine learning, ML, model / functionality, a report for one or both of an inference and a prediction, the report comprising an indication that the one or both of the inference and the prediction is one or more of invalid, out of range, and inaccurate; andreceiving (S202) the report from the UE.

8. The method of claim 7, wherein the report comprises radio measurement predictions of measurement resources that comprise one or more of:prediction results for at least one measurement quantity comprising one or more of layer 1 reference signal received power, reference signal received quality, RSRQ, layer 1 signal to interference noise ratio, SINR, and received signal strength indicator, RSSI; prediction results for one or more best beam predictions; andradio measurement prediction results for one or more occasions in time.

9. The method of claims 7 or 8, wherein the indication indicates the one or both of the inference and the prediction is one or more of invalid, out of range, and inaccurate is based on one or more of:at least one applicability condition of the ML model / functionality; performance of the ML model / functionality;inability of the ML model / functionality to be executed at the UE;an inaccuracy in the report;the report comprising a predicted quantity that is out of range;measurement inputs to the ML model / functionality being fewer than a predetermined quantity; andthe UE requesting deactivation of the ML model / functionality and not receiving an indication from the network node to deactivate the ML model / functionality.

10. The method of any one of claims 7-9, wherein the indication comprises one or more of:a request from the UE to deactivate the ML model / functionality; anda reason the report is at least one of invalid and out of range.

11. The method of any one of claim 7-10, wherein the method comprises transmitting another indication to the UE to deactivate the ML model / functionality.

12. The method of any one of Claims 7-11, further comprising:receiving (S203) an indication from the UE, indicating a reason the report is one or more of invalid, out of range, and inaccurate.

13. The method of Claim 12, wherein the indication received from the UE is received after the report.

14. A user equipment, UE (22), configured to communicate with a network node (16), the UE (22) comprising at least processing circuitry that is configured to perform a method according to any one of Claims 1-6.

15. A network node (16) configured to communicate with user equipment, UE (22), the network node (16) comprising at least processing circuitry that is configured to perform a method according to any one of Claims 7-13.

Images