Method, apparatus and computer program
By enabling user equipment to report intermediate key performance indicators in configured formats, the solution addresses the challenge of managing machine learning models in communication systems, improving channel estimation and prediction accuracy.
Patent Information
- Application Number
- PCT/EP2026/063211
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- Priority Date
- 2025-05-09
- Filing Date
- 2026-05-07
- Publication Date
- 2026-11-12
AI Technical Summary
Existing communication systems lack efficient methods for reporting intermediate key performance indicator values from user equipment to network access nodes, particularly in the context of machine learning models used for channel estimation, which hinders effective life cycle management of these models.
A user equipment deploys a machine learning model to estimate communication channel characteristics and transmits capability information to a network access node, allowing the node to select a format for reporting intermediate key performance indicator values, such as squared generalized cosine similarity, based on configured parameters, enabling accurate and efficient reporting.
This approach allows network access nodes to effectively manage the life cycle of machine learning models by receiving reliable and precise intermediate key performance indicator values, enhancing the accuracy and efficiency of channel estimation and prediction.
Smart Images

Figure EP2026063211_12112026_PF_FP_ABST
Abstract
Description
METHOD, APPARATUS AND COMPUTER PROGRAMFIELD
[0001] The present application relates to reporting intermediate key performance indicator values.BACKGROUND
[0002] A communication system can be seen as a facility that enables communication sessions between two or more entities such as user terminals, base stations and / or other nodes by providing carriers between the various entities involved in the communications session. A communication system can be provided for example by means of a communication network and one or more compatible communication devices. The communication sessions may comprise, for example, communication of data for carrying communications such as voice, video, electronic mail (email), text message, multimedia and / or content data and so on. Non-limiting examples of services provided comprise two-way or multi-way calls, data communication or multimedia services and access to a data network system, such as the Internet.
[0003] The communication system and associated devices typically operate in accordance with a given standard or specification which sets out what the various entities associated with the system are permitted to do and how that should be achieved. Communication protocols and / or parameters which shall be used for the connection are also typically defined. One example of a communications system is UTRAN (Universal Mobile Telecommunications Service terrestrial radio access network (e.g., 3G radio)). Other examples of communication systems are the longterm evolution (LTE) of the Universal Mobile Telecommunications System (UMTS) radio-access technology and so-called 5G or New Radio (NR) networks. NR is being standardized by the 3rd Generation Partnership Project (3GPP).SUMMARY
[0004] According to a first aspect, there is provided an apparatus comprising: at least one processor; and at least one memory comprising code that, when executed by the at least one processor, causes the apparatus to perform: deploying, at the apparatus, a machine learning model for estimating one or more characteristics of a communication channel; transmitting, to a network access node, capability information, wherein the capability information indicates one or more formats that maybe used by the apparatus for transmitting a report comprising information indicating one or more intermediate key performance indicator values of the machine learning model; receiving, from the network access node, a configuration for reporting the information indicating one or more intermediate key performance indicator values of the machine learning model; selecting the a first format from the one or more formats for reporting the information indicating the one or more intermediate key performance indicator values to the network access node based on the received configuration; and transmitting the report comprising the information indicating the one or more intermediate key performance indicator values to the network access node in accordance with the first format.
[0005] According to a second aspect, there is provided an apparatus comprising means for performing: deploying, at the apparatus, a machine learning model for estimating one or more characteristics of a communication channel; transmitting, to a network access node, capability information, wherein the capability information indicates one or more formats that may be used by the apparatus for transmitting a report comprising information indicating one or more intermediate key performance indicator values of the machine learning model; receiving, from the network access node, a configuration for reporting the information indicating one or more intermediate key performance indicator values of the machine learning model; selecting a first format from the one or more formats for reporting the information indicating the one or more intermediate key performance indicator values to the network access node based on the received configuration; and transmitting the report comprising the information indicating the one or more intermediate key performance indicator values to the network access node in accordance with the first format.
[0006] According to a third aspect, there is provided a method for an apparatus, the method comprising: deploying, at the apparatus, a machine learning model for estimating one or more characteristics of a communication channel; transmitting, to a network access node, capability information, wherein the capability information indicates one or more formats that may be used by the apparatus for transmitting a report comprising information indicating one or more intermediate key performance indicator values of the machine learning model; receiving, from the network access node, a configuration for reporting the information indicating one or more intermediate key performance indicator values of the machine learning model; selecting a first format from the one or more formats for reporting the information indicating the one ormore intermediate key performance indicator values to the network access node based on the received configuration; and transmitting the report comprising the information indicating the one or more intermediate key performance indicator values to the network access node in accordance with the first format.
[0007] According to a fourth aspect, there is provided an apparatus comprising: deploying circuitry for deploying, at the apparatus, a machine learning model for estimating one or more characteristics of a communication channel; transmitting circuitry for transmitting, to a network access node, capability information, wherein the capability information indicates one or more formats that may be used by the apparatus for transmitting a report comprising information indicating one or more intermediate key performance indicator values of the machine learning model; receiving circuitry for receiving, from the network access node, a configuration for reporting the information indicating one or more intermediate key performance indicator values of the machine learning model; selecting circuitry for selecting a first format from the one or more formats for reporting the information indicating the one or more intermediate key performance indicator values to the network access node based on the received configuration; and transmitting circuitry for transmitting the report comprising the information indicating the one or more intermediate key performance indicator values to the network access node in accordance with the first format.
[0008] The following may apply in respect of any (e.g., one or more, including all) of the above first to fourth aspects.
[0009] The transmitting the report comprising the information indicating the one or more intermediate key performance indicator values to the network access node in accordance with the first format may comprise at least one of: transmitting a report comprising information indicating the one or more intermediate key performance indicator values per frequency sub-band, per layer, or per prediction time instance of the machine learning model.
[0010] According to a fifth aspect, there is provided an apparatus comprising: at least one processor; and at least one memory comprising code that, when executed by the at least one processor, causes the apparatus to perform: receiving, from a user equipment, capability information, wherein the capability information indicates one or more formats that may be used by the user equipment for transmitting a report comprising information indicating one or more intermediate key performance indicatorvalues for estimating one or more characteristics of a communication channel of a machine learning model deployed at the user equipment; determining, based on the capability information, a configuration for reporting the information indicating the one or more intermediate key performance indicator values; transmitting, to a user equipment, information indicating the configuration; and receiving, from the user equipment, the report comprising the information indicating the one or more intermediate key performance indicator values in a first format, the first format being one of the one or more formats.
[0011] According to a sixth aspect, there is provided an apparatus comprising means for performing: receiving, from a user equipment, capability information, wherein the capability information indicates one or more formats that may be used by the user equipment for transmitting a report comprising information indicating one or more intermediate key performance indicator values for estimating one or more characteristics of a communication channel of a machine learning model deployed at the user equipment; determining, based on the capability information, a configuration for reporting the information indicating the one or more intermediate key performance indicator values; transmitting, to a user equipment, information indicating the configuration; and receiving, from the user equipment, the report comprising the information indicating the one or more intermediate key performance indicator values in a first format, wherein the first format being one of the one or more formats.
[0012] According to a seventh aspect, there is provided a method for an apparatus, the method comprising: receiving, from a user equipment, capability information, wherein the capability information indicates one or more formats that may be used by the user equipment for transmitting a report comprising information indicating one or more intermediate key performance indicator values for estimating one or more characteristics of a communication channel of a machine learning model deployed at the user equipment, wherein the one or more formats comprises a first format; determining, based on the capability information, a configuration for reporting the information indicating the one or more intermediate key performance indicator values; transmitting, to a user equipment, information indicating the configuration; and receiving, from the user equipment, the report comprising the information indicating the one or more intermediate key performance indicator values in a first format, the first format being one of the one or more formats.
[0013] According to an eighth aspect, there is provided an apparatus comprising: receiving circuitry for receiving, from a user equipment, capability information, wherein the capability information indicates one or more formats that may be used by the user equipment for transmitting a report comprising information indicating one or more intermediate key performance indicator values for estimating one or more characteristics of a communication channel of a machine learning model deployed at the user equipment; determining circuitry for determining, based on the capability information, a configuration for reporting the information indicating the one or more intermediate key performance indicator values; transmitting circuitry for transmitting, to the user equipment, information indicating the configuration; and receiving circuitry for receiving, from the user equipment, the report comprising the information indicating the one or more intermediate key performance indicator values t in a first format, the first format being one of the one or more formats.
[0014] The following may apply in respect of any (e.g., one or more, including all) of the above fifth to eighth aspects.
[0015] The receiving the report comprising the information indicating the one or more intermediate key performance indicator values in the first format may comprise at least one of: receiving a report comprising information indicating the one or more intermediate key performance indicator values per frequency sub-band, per layer, or per prediction time instance of the machine learning model.
[0016] The following may apply in respect of any (e.g., one or more, including all) of the above first to eighth aspects.
[0017] The one or more intermediate key performance indicator values may comprise at least one of: a squared generalized cosine similarity value, or a value representing a plurality of squared generalized cosine similarity values.
[0018] The information indicating the one or more intermediate key performance indicator values may indicate, for each of one or more determined intermediate key performance indicator value, a difference between that determined intermediate key performance indicator value and a respective reference intermediate key performance indicator value, and the configuration defines a range of said differences that are to be reported via the first format.
[0019] The first format may comprise a plurality of bits, each of said bits corresponding to at least one of: a first subrange of values of said range, a second subrange of values of said range, a first out-of-range indication for a difference value that is above theupper limit of the range, or a second out-of-range indication for a difference value that is below the lower limit of the range.
[0020] The first subrange may define a first subrange of values, and the second subrange may define a second subrange of values. The first and the second subrange of values may be equally sized. The first and second subrange of values may be unequally sized.
[0021] The reference intermediate key performance indicator value may be based on a last channel observation obtained relative to a ground truth determined at a prediction time instance
[0022] The one or more intermediate key performance indicator values may comprise a value representing a plurality of squared generalized cosine similarity values by comprising at least one of: a mean or median value of a plurality of determined intermediate key performance indicator values or a mean or median value of, for each of the plurality of determined intermediate key performance indicator values, a difference between that determined intermediate key performance indicator value and a respective reference intermediate key performance indicator value.
[0023] The report may further comprise information indicating a reliability of the one or more intermediate key performance indicator values.
[0024] The information indicating the reliability may comprise at least one of: a variance of the one or more intermediate key performance indicator values, a signal to interference and noise ratio, ora delay spread. The delay spread may be considered to be a measure of how far the delays of the multipath components comprising a radio channel vary statistically.
[0025] The report may comprise information indicating one or more intermediate key performance indicator values for a plurality of at least one of the following domains: a frequency domain, a spatial domain, or a time domain.
[0026] The one or more intermediate key performance indicator values comprised in the report may be, for each of the one or more intermediate key performance indictor values, weighted by a reliability of that intermediate key performance indicator value.
[0027] The report may comprise an identifier of a codebook value, wherein the identifier of the codebook value indicates the one or more intermediate key performance indictor values across at least one of the frequency domain, spatial domain, and time domain.
[0028] The identifier of the codebook value may indicate the one or more intermediate key performance indictor values across a plurality of the frequency domain, spatial domain, and time domain.
[0029] The identifier of the codebook may identify a respective relative value of at least one of the frequency domain, spatial domain, or time domain, wherein a relative value indicates a difference between the value being reported and a corresponding strongest intermediate key performance indictor value.
[0030] The one or more intermediate key performance indicator values may comprise a multi-instance report. For example, the one or more intermediate key performance indicator values may comprise multiple statistical values, each of the multiple statistical values being associated with a respective error correction value, and each of the multiple statistical values corresponding to a respective monitoring instance.
[0031] The apparatus of the fifth to eighth aspects may receive another report from the user equipment in accordance with the first format, the another report comprising respective information indicating one or more corresponding intermediate key performance indicator values, wherein the one or more corresponding intermediate key performance indicator values comprises multiple corresponding statistical values, each of the multiple corresponding statistical values being associated with a respective error correction value that is based on respective error correction value(s) comprised in the report, and perform an averaging operation across at least values comprised in the report and the another report to obtain a mean key performance indicator value.
[0032] The apparatus of the first to fourth aspects may transmit, to the network access node, another report in accordance with the first format, the another report comprising respective information indicating one or more corresponding intermediate key performance indicator values, wherein the one or more corresponding intermediate key performance indicator values comprises multiple corresponding statistical values, each of the multiple corresponding statistical values being associated with a respective error correction value that is based on respective error correction value(s) comprised in the report.
[0033] According to an aspect, there is provided a non-transitory computer readable medium comprising program instructions that, when executed by an apparatus, cause the apparatus to perform at least the method according to any of the preceding aspects.
[0034] In the above, many different embodiments have been described. It should be appreciated that further embodiments may be provided by the combination of any two or more of the embodiments described above.DESCRIPTION OF FIGURES
[0035] Embodiments will now be described, by way of example only, with reference to the accompanying Figures in which:
[0036] Figure 1 shows a representation of a network system according to some example embodiments;
[0037] Figure 2 shows a representation of a control apparatus according to some example embodiments;
[0038] Figure 3 shows a representation of an apparatus according to some example embodiments;
[0039] Figure 4 illustrates example signalling that may be performed by apparatus described herein;
[0040] Figure 5 illustrates example differences between a machine learning model intermediate key performance indicator value and a reference intermediate key performance indicator value as a cumulative distribution function;
[0041] Figure 6 illustrates how accuracy of an intermediate key performance indicator value may value across different spatial layers;
[0042] Figure 7 illustrates an example codebook;
[0043] Figures 8A and 8B illustrate example intermediate key performance indicator values across different user equipment and / or across different time instances; and
[0044] Figures 9 to 10 illustrate example methods that may be performed by apparatus described herein.DETAILED DESCRIPTION
[0045] The following describes operations that may be performed in relation to a network access node obtaining information relating to one or more intermediate key performance indicator (KPI) values of a machine learning model deployed at a user equipment.
[0046] An intermediate KPI, sometimes called a proxy metric, is a key performance indicator that's used to measure progress towards a larger, overarching goal. The network access node may use the obtained KPI values to determine whether at leastone setting (e.g., a parameter and / or condition) of the machine learning model should be changed in order that more accurate predictions are determined by that machine learning model.
[0047] The described machine learning model may be used for any of a plurality of different predictions. In an implementation example, the machine learning model deployed at the user equipment may estimate a channel quality based on received channel state information transmissions made by a network access node.
[0048] Although these, and further features, will be elaborated on below, Figures 1 to 3 are first provided to illustrate an example communication environment in which the presently described methods may be performed.
[0049] Figure 1 shows an example communication environment 100 in which example embodiments of the present disclosure can be implemented.
[0050] In the communication environment 100, a plurality of communication devices, comprising user devices 110 and 115 (also referred to herein as a “terminal” or “terminal device”) and a network device 120 (also referred to herein as a “network access node”), can communicate with each other. The network device 120 may serve a coverage area, called a cell 125. The user device 110 may have access to a communication network via the cell 125. In some example embodiments, both the user device 110 and the network device 120 may be configured to implement a beamforming technique and communicate with each other via a plurality of beams.
[0051] The term “terminal device” refers to any end device that may be capable of wireless communication. By way of example rather than limitation, a terminal device may also be referred to as a communication device, user equipment (UE), a Subscriber Station (SS), a Portable Subscriber Station, a mobile device, a Mobile Station (MS), or an Access Terminal (AT). The terminal device may include, but not limited to, a mobile phone, a cellular phone, a smart phone, voice over IP (VoIP) phones, wireless local loop phones, a tablet, a wearable terminal device, a personal digital assistant (PDA), portable computers, desktop computer, image capture terminal devices such as digital cameras, gaming terminal devices, music storage and playback appliances, vehicle-mounted wireless terminal devices, wireless endpoints, mobile stations, laptop-embedded equipment (LEE), laptop-mounted equipment (LME), USB dongles, smart devices, wireless customer-premises equipment (CPE), a machine-type communications (MTC) device, an Internet of Things (loT) device, a watch or other wearable, a head-mounted display (HMD), a vehicle, a drone, a medicaldevice and applications (e.g., remote surgery), an industrial device and applications (e.g., a robot and / or other wireless devices operating in an industrial and / or an automated processing chain contexts), a consumer electronics device, a device operating on commercial and / or industrial wireless networks, and the like. The terminal device may also correspond to a Mobile Termination (MT) part of an IAB node (e.g., a relay node). In the following description, the terms “terminal device”, “communication device”, “terminal”, “user device”, “user equipment” and “UE” may be used interchangeably.
[0052] As used herein, the term “network device” is used interchangeably with “network access node”, and refers to a node in a communication network via which a terminal device accesses the network and receives services therefrom. The network device may refer to a base station (BS) or an access point (AP), for example, a node B (Node or NB), an evolved NodeB (eNodeB or eNB), an NR NB (also referred to as a gNB), a Remote Radio Unit (RRU), a radio header (RH), a remote radio head (RRH), a relay, an Integrated Access and Backhaul (IAB) node, a low power node such as a femto, a pico, a non-terrestrial network (NTN) or non-ground network device such as a satellite network device, a low earth orbit (LEO) satellite and a geosynchronous earth orbit (GEO) satellite, an aircraft network device, and so forth, depending on the applied terminology and technology. In some example embodiments, radio access network (RAN) split architecture comprises a Centralized Unit (CU) and a Distributed Unit (DU) at an IAB donor node. An IAB node comprises a Mobile Terminal (IAB-MT) part that behaves like a UE toward the parent node, and a DU part of an IAB node behaves like a base station toward the next-hop IAB node.
[0053] In some example embodiments, a link from the network device 120 to the user device 110 or 115 is referred to as a DL, while a link from the user device 110 or 115 to the network device 120 is referred to as a UL. Links are also referred to herein as “channels”. In DL, the network device 120 is a Tx device (or a transmitter), and the user device 110 or 115 is a Rx device (or a receiver). In UL, the user device 110 or 115 is a Tx device (or a transmitter), and the network device 120 is a Rx device (or a receiver). A link between the user device 110 and another user device (not shown) is referred to as a sidelink (SL). In SL, one of the user devices is a Tx device (or a transmitter), and the other of the user devices is a Rx device (or a receiver).
[0054] Communications in the communication environment 100 may be implemented according to any proper communication protocol(s), comprising, but not limited to,cellular communication protocols of the first generation (1G), the second generation (2G), the third generation (3G), the fourth generation (4G), the fifth generation (5G), the sixth generation (6G), and the like, wireless local network communication protocols such as Institute for Electrical and Electronics Engineers (IEEE) 802.11 and the like, and / or any other protocols currently known or to be developed in the future. Moreover, the communication may utilize any proper wireless communication technology, comprising but not limited to: Code Division Multiple Access (CDMA), Frequency Division Multiple Access (FDMA), Time Division Multiple Access (TDMA), Frequency Division Duplex (FDD), Time Division Duplex (TDD), Multiple-Input Multiple-Output (MIMO), Orthogonal Frequency Division Multiple (OFDM), Discrete Fourier Transform spread OFDM (DFT-s-OFDM) and / or any other technologies currently known or to be developed in the future.
[0055] Figure 2 illustrates an example of a control apparatus 200 for causing a network device 120 (such as the network device described in Figure 1) to perform its operations. The control apparatus may comprise at least one random access memory (RAM) 211a, at least on read only memory (ROM) 211b, at least one processor 212, 213 and an input / output interface 214. The at least one processor 212, 213 may be coupled to the RAM 211a and the ROM 211b. The at least one processor 212, 213 may be configured to execute an appropriate software code 215. The software code 215 may for example allow to perform one or more steps to perform one or more of the present aspects. The software code 215 may be stored in the ROM 211b. The control apparatus 200 may be interconnected with another control apparatus 200 controlling another function of the network device. In some embodiments, each function of the network device comprises a control apparatus 200. In some exemplary embodiments, the apparatus 200 may be implemented at the network device 120 or may be the network device 120.
[0056] Figure 3 illustrates an example of a terminal 300, such as the user device 110, 115 illustrated on Figure 1. The terminal 300 may be provided by any device capable of sending and receiving radio signals, such as the user device described herein. The terminal 300 may provide, for example, communication of data for carrying communications. The communications may be one or more of voice, electronic mail (email), text message, multimedia, data, machine data and so on.
[0057] The terminal 300 may receive signals over an air or radio interface 307 via appropriate apparatus for receiving and may transmit signals via appropriateapparatus for transmitting radio signals. In Figure 3 transceiver apparatus is designated schematically by block 306. The transceiver apparatus 306 may be provided for example by means of a radio part and associated antenna arrangement. The antenna arrangement may be arranged internally or externally to the mobile device.
[0058] The terminal 300 may be provided with at least one processor 301 , at least one memory ROM 302a, at least one RAM 302b and other possible components 303 for use in software and hardware aided execution of tasks it is designed to perform, including control of access to and communications with access systems (such as a network access system provided by the network device described above in relation to Figures 1 and 2) and other communication devices. The at least one processor 301 is coupled to the RAM 302b and the ROM 302a. The at least one processor 301 may be configured to execute an appropriate software code 308. The software code 308 may for example allow to perform one or more of the present aspects. The software code 308 may be stored in the ROM 302a.
[0059] The processor, storage and other relevant control apparatus can be provided on an appropriate circuit board and / or in chipsets. This feature is denoted by reference 304. The device may optionally have a user interface such as key pad 305, touch sensitive screen or pad, combinations thereof or the like. Optionally one or more of a display, a speaker and a microphone may be provided depending on the type of the device.
[0060] In some exemplary embodiments, the terminal 300 may be an apparatus comprising at least one processor and at least one memory storing instructions that, when executed by the at least one processor, cause a user device 110, 115 to perform examples or embodiments described in this document.
[0061] In communication systems such as those of Figures 1 to 3, the UE may be configured to perform at least one channel prediction method.
[0062] Channel prediction relates to estimating (e.g., forecasting or predicting) the characteristics that a wireless channel is likely to have in the future.
[0063] In Rel. 18, channel estimation is performed using methods deployed on the physical (PHY) layer. However, it is expected that Rel. 19 and beyond may enable cases in which artificial intelligence (Al) and / or machine learning (ML) can be used to perform at least one channel prediction method. However, in contrast to the physicallayer methods, it would be useful to include some method for monitoring how well an AI / ML model works under different certain channel conditions.
[0064] The following aims to address one or more of the above-mentioned issues.
[0065] In particular, the following describes one or more methods that may be used for reporting intermediate key performance indicator values from a user equipment to a network access node.
[0066] In examples described in the following, a user equipment may provide a network access node with capability information that indicates one or more formats that may be used by the user equipment for transmitting a report comprising information indicating one or more intermediate key performance indicator values of a machine learning model deployed at the user equipment. The network access node may use this capability information to select a first format from the plurality of formats for use by the UE in transmitting the report, and provide information identifying this first format to the UE. The UE may subsequently transmit the report comprising information indicating one or more intermediate key performance indicator values of the machine learning model in accordance with the first format.
[0067] The following illustrates these principles via one or more implementation examples.
[0068] Figure 4 illustrates an implementation example in which a user equipment 401 and a network access node 402 each perform one or more methods.
[0069] During 4001, the user equipment 401 signals capability information to the network access node 402. This capability information may provide information to the network access node that indicates which reporting formats can be used by the user equipment for reporting intermediate key performance indicators to the network access node. Although the capability information may be provided at any time, the UE may provide this capability information to the network access node when registering with the network access node.
[0070] Example intermediate key performance indicators may comprise at least one of a normalized mean square error (NMSE) or a squared generalized cosine similarity (SGCS). An NMSE is a statistical measure that may be used to quantify the difference between a model's predictions and the true values. It's a type of "error" metric, with lower values indicating better model performance. A squared generalized cosine similarity is a variant of a standard cosine similarity that has previously been considered in 3GPP.
[0071] It is understood in the following that any type of intermediate key performance indicator type may be used in the presently described methods. For clarity and brevity, the following will consider intermediate KPIs in the context of SGCS values. However, it is understood that this is not limiting.
[0072] A report format may define what information is to be reported, and how that information is to be expressed (e.g., comprised) in the report. An indication of example report formats is provided at the end of the discussion of Figure 4.
[0073] During 4002, the network access node 402 selects a report format for reporting intermediate key performance indicator values based on the capability information signalled during 4001. The network access node may further select one or more other parameters that are associated with the report (although it is understood that in some examples, this information may also be defined as part of a report format). For example, the network access node may define a length of a monitoring instance (e.g., a duration of time over which a key performance indicator value is determined) and / or a frequency of reporting one or more intermediate KPI values from the UE to the network access node. The network access node may define one or more of these parameters at a radio resource control protocol level.
[0074] During 4003, the network access node 401 signals the UE 401. This signalling may comprise an indication of the selecting report format of 4002. This signalling may comprise an indication of the one or more other parameters selected by the network access node during 4002.
[0075] During 4004, the network access node 402 transmits at least one channel state information (CSI) reference signals that are received by the UE 401.
[0076] Channel state information (CSI) is obtained by the UE based on the reference CSI reference signals transmitted by the network access node. CSI may comprise channel feedback parameters such as, for example, channel quality indicator (CQI), rank indicator (Rl), and precoding matrix indicator (PMI). When these channel feedback parameters are provided to the network access node, they may be used by the network access node for downlink precoding. The CSI may be considered to comprise estimated channel characteristics.
[0077] For those CSI reference signals that are received by the UE 401 , the UE may perform at least one of 4005 to 4007 and 4008 when a channel estimation is performed for the received CSI reference signals using a machine learning model deployed at the UE.
[0078] During 4005, the UE 401 estimates intermediate key performance indicator values of the machine learning model deployed at the UE. In an example, the intermediate key performance indicator values may be an SGCS for the combination of the machine learning model prediction and a reference prediction value. The reference prediction value may be, for example, a previous output of the machine learning model prediction when the reference predictor (e.g., the entity that outputs the reference prediction value) is a zero order hold apparatus. The intermediate key performance indicator values may be determined (e.g., measured and / or calculated) across at least one of a frequency domain (e.g., one or more frequency sub-bands), a time domain (e.g., one or more prediction time instances), or a spatial domain (e.g., one or more layers).
[0079] During 4006, the UE 401 calculates a weighted sum of the estimated intermediate key performance indicator values over at least one of a frequency domain, a time domain, or a spatial domain. The UE 401 may use this weighted sum to determine a respective mean value of the estimated intermediate key performance indicator values for one or more of the domains. The respective mean value may be used for determining a value representing a range of intermediate key performance indicators values to be reported (e.g., to determine a reporting window length) and / or lengths of sub-ranges of this range. This is further discussed below in the examples of the one or more formats.
[0080] During 4007, the UE 401 forms a report comprising information relating to multiple estimated intermediate key performance indicator values, each of the multiple estimated intermediate key performance indicator values corresponding to a respective monitoring instance. Stated differently, during 4007, the UE 401 forms a multi-instance report comprising error detection, as described further below.
[0081] During 4008, the UE 401 transmits, to the network access node 402, a report comprising information indicating one or more intermediate key performance indicator values. The report of 4008 may be formatted as per the format indicated in the signalling of 4003.
[0082] During 4009, the network access node 402 uses the report of 4008 (and any other reports received from the UE that provides information indicating one or more intermediate key performance indicator values) to determine whether one or more parameters of the machine learning model deployed at the UE should change. Stateddifferently, during 4009, the network access node makes a decision relating to a life cycle management of the machine learning model is made.
[0083] During 4010, assuming that the network access node determined during 4009 that at least part of an operation of the machine learning model should change, the network access node signals the UE 401 to cause the UE to implement the change. This can be, for example, an indirect signaling by reconfiguring the CSI reference signal configuration as discussed by 3GPP Rel19 RAN2 for AI / ML life cycle management (LCM) for different sub use cases (e.g., beam management, channel prediction, etc.).
[0084] The following illustrates one of more example formats that the report(s) from the UE may take. It is understood that these examples are not mutually exclusive, and that different features of these methods may be combined together in a plurality of different ways.
[0085] In a first example, the one or more format comprises a report comprising bits that indicate whether an intermediate KPI value being reported falls within a predetermined range of values.
[0086] For example, when the KPI value being reported is an SGCS value, the report may comprise indications of SGCS values relative to a reference predicted value, there the report is limited to a range of the SGCS values.
[0087] Reporting of relative SGCS values between reference and AI / ML predictor leads to a significantly smaller value range like [-0.1 ...0.2] compared to the absolute SGCS value range of [0 ...1]. Therefore, assuming the same number of bits, a significantly higher accuracy for the difference in SGCS values can be achieved when the reporting of an intermediate KPI value is performed with reference to a reference value (e.g., the previously determined intermediate KPI value when a zero order hold method is used to determine the reference KPI value). This accuracy may be improved when this technique is combined with a non-linear quantization of the reporting bits in the bit map such that at least two bits in the bitmap correspond to different subrange sizes to each other (as illustrated further below).
[0088] Further, the limited range of values may be reported such that each bit position in the report corresponds to a sub-range of the range of values (e.g., less than all), and / or to an indication that the intermediate value being reported is outside of the range.
[0089] The subranges that, when combined, make up to the limited range, may be non-liner (e.g., a first bit may indicate a first subrange of 0.0 to 0.1 , a second subrange may indicate a subrange of 0.1 to 0.125, etc.). Stated differently, the report of the first example may comprise a fixed bit-field (e.g., 4 bits), where each bit and / or combination of bits corresponds to a different range of intermediate KPI values.
[0090] As another example, an intermediate KPI value as 0.5 to 0.6 (e.g., entry ‘0001’), and another entry may represent a value subrange for SGCS as 0.6 to 0.65 (e.g. ,’001 O’), where subranges represented by two entries are not the same.
[0091] A non-linear quantization of subranges within the window of relevant SGCS values may account for an expected exponential shape of the cumulative distribution function (CDF) of the intermediate key performance indicators (e.g., of the SGCS CDF). A CDF may be considered to be a function that describes the probability that a random variable will take a value less than or equal to a specified value.
[0092] A mapping between each bit value and their respective subranges may be stored at a UE using a predefined table. Such a table may be useful for reconfiguring the subranges and / or the limited range flexibly, based on current reported intermediate KPI values. The window of values between the highest and lowest value might then be quantized linearly or non-linearly. The table, as well as the upper and lower value of the reporting window, may be configured by a network access node. The network access node may further update this table when different sub-ranges and / or a different limited range is to be reported.
[0093] For example, the network access node may cause the UE to be configured with the limited range itself, with the UE determining what subrange corresponds to which bit in the bitfield by applying a predetermined algorithm to the limited range. As another example, the network access node may be caused to configure at the UE both the limited range and what subrange corresponds to which bit in the bitfield.
[0094] The configuration may be performed when the network access node indicates to the UE that the UE is to report an intermediate KPI value using a specific format. However, it is understood that this configuration may be performed at any time prior to the intermediate KPI value being reported.
[0095] At least one entry of the fixed bit field may further represent an indication that the intermediate KPI value is out of range of the limited range of intermediate KPI values being reported. In one example, entry “0000” (assuming a case of 4-bitintermediate KPI value reporting) may be used for indicating that the intermediate KPI value being reported is outside of the limited range.
[0096] There may be at least two bits of the bit field that indicate that the intermediate KPI value is outside of the limited range being reported. For example, there may be provided a first bit in the bit field that indicates that the intermediate KPI value being reported is higher than the highest value defined by the limited range. There may further be provided a second bit in the bit field that indicates that the intermediate KPI value being reported is lower than the lowest value defined by the limited range.
[0097] Applying a window of for reporting specific key performance indicator values and ‘out of range’ values may focus the network access node to evaluate the performance of the AI / ML predictor model deployed at the UE where the evaluation may have the greatest effect. For example, very low and extremely high SGCS values (e.g., where the predictor either fails completely or performs almost perfectly) are ignored.
[0098] It is understood that the size and / or starting and end points of at least one of the limited range and sub-ranges may be reconfigured based on a number of intermediate KPI values being reported as being outside of the limited range. For example, (e.g., lower SGCS values compared to lower limit of the window of relevant SGCS values).
[0099] In a second example of the form the format may take, the UE is configured to perform a reporting of the intermediate KPI value with respect to at least one of a spatial domain, a time domain, or a frequency domain.
[0100] For example, the reporting of the intermediate KPI value may be configured to report at least one of per spatial domain, per time domain, or per frequency domain. Stated differently, the reporting of the KPI value may be configured to be performed at least one of per layer, per prediction time instance, or per frequency sub-band.
[0101] As another example, an intermediate KPI value may be reported as an average across multiple domains, the format may indicate that an intermediate KPI value is to be reported in the form of an average across multiple layers (e.g., across multiple instances of a spatial domain), across multiple prediction instances (e.g., across multiple instances of a time domain), and / or across multiple frequency sub-bands (e.g., across multiple instances of a frequency domain).
[0102] In a third example of the form the format may take, the UE may be configured to report both a first value of an intermediate KPI value (e.g., a first layer SGCS value)using a first bit field, and a second value of an intermediate KPI (e.g., a second, third, and / or fourth layer SGCS value) using a second bit field to report a differential SGCS.
[0103] This third example is illustrated with respect to Figure 5, which illustrates cumulative distribution functions (CDFs) of a reference intermediate KPI value (represented by the zero order hold (ZOH) plot, which is the upper plot on the graph) and an artificial intelligence (Al) and / or machine learning (ML) predictor intermediate KPI value (represented by the LSTM plot, which is lower on the graph).
[0104] In this example, the network access node may limit the reporting to be performed (e.g., by selecting a specific format) so that only those values falling between 0.8 and 0.97 of the cosine similarity squared metric are reported by the UE. This may be because, for example, the accuracy of the prediction reduces for lower SGCS values the prediction fails completely. Using the CDFs as illustrated in Figure 5 may result in a report of a single intermediate KPI value, such as a single SGCS value corresponding to the intermediate KPI value at 50 % of the CDF plot.
[0105] It is understood that, as an alternative to CDFs, the format of the reporting may be based on at least one of a probability density function (pdfs) or a histogram. A pdf may be considered to be a function whose value at any given sample (or point) in the sample space (the set of possible values taken by the random variable) can be interpreted as providing a relative likelihood that the value of the random variable would be equal to that sample. A histogram may be considered to be a chart that plots the distribution of a numeric variable's values as a series of bars. Each bar typically covers a range of numeric values called a bin or class. A bar's height indicates the frequency of data points with a value within the corresponding bin.
[0106] When a pdf is used for reporting, a mean value of the intermediate KPI value may be useful.
[0107] When histograms are used, there may be a strong deviation of the histograms from a typical Gaussian distribution. The format may therefore be configured to additionally report one or more statistics of the intermediate KPI value(s). For example, the one or more statistics may comprise a mean value for an AI / ML predictor intermediate KPI value output and a mean value fora reference predictor intermediate KPI value (e.g., the zero order hold mentioned herein). The format may further comprise information indicating a respective reliability of the reported intermediate KPI value(s) and / or their mean. The information indicating the respective reliability may becalculated by the UE based on a determined variance of the intermediate KPI values, a signal to interference and noise ratio (SNR), a delay spread, etc.
[0108] The reliability of the intermediate KPI values, such as SGCS values, may vary significantly over different domains (f, t and I) as well as over different monitoring time instances. Adding corresponding reliability information to the information to be reported by the UE (e.g. derived from the variance, SINR, channel delay spread and / or others) may therefore provide a network access node with information for combining multiple monitoring reports before making a life cycle management decision about the ML model deployed at the UE.
[0109] Reporting of mean values may be advantageous in that the mean values may combine the mean values of multiple reporting instances to report a combined mean value over a plurality of instances. Further, reporting of mean values may combine the mean values of multiple frequency sub-bands f, multiple prediction time instances t, and / or multiple layers I.
[0110] When the format indicates that the UE is to report one or more statistical values at a granularity that reflects at least one of a spatial domain, time domain, and / or frequency domain (such as, for example, reporting a mean intermediate KPI value individually per frequency sub-band, per prediction time instance, and / or per layer), then a very efficient reporting of the mean values would be useful.
[0111] It is understood that combinations of the above-mentioned features may be performed.
[0112] For example, while relative values between the intermediate KPI value(s) of a reference predictor (e.g., a zero order hold, ZOH, process) and the intermediate KPI value(s) of the ML model may be usefully reported, it may be further useful to format the report to comprise an absolute mean value (for example of the reference predictor) so that the network access node knows an average intermediate KPI value as well as a relative gain of an intermediate KPI value. For example, when the average intermediate KPI value is close to 0.5, an intermediate KPI value increase of 0.05 has more of an effect than the same increase would have on an intermediate KPI value of 0.9. For example.
[0113] As a fourth example, the format to be used for reporting may be configured to comprise statistical intermediate KPI values of a plurality of monitoring instances. For example, one monitoring instance may not be sufficient for enabling the network access node to make a decision about whether the machine learning model deployedat the UE is producing accurate CSI predictions, and so the network access node may be configured to wait to make a decision about the ML model until multiple reports have been received.
[0114] To streamline this process, the network access node may define a format for reporting that defines (e.g., sets) a number N of monitoring instances to be reported in a single report. Such a report is referred to herein as a multi-instance report. The multi-instance report may comprise inherent error correction.
[0115] For example, when a single monitoring instance is expected to be associated with a single statistical value to represent an intermediate KPI value (e.g., a mean value) with a coarse quantisation, the following reporting instances may take reporting errors from previous reporting instances into account when calculating a current error value (e.g., a mean value of the reporting errors). The reported error (e.g., the error value comprised in the report) may comprise a mean value relative to the previously reported error values. After the network access node has collected all N reported intermediate KPI values (which may be comprised in a single report and / or in multiple reports), the network access node sums the reported intermediate KPI values from the N reporting instances and divides it by the number of reporting instances to obtain a refined combined mean value. The effective quantization error of the combined mean value is smaller than that of an individual reporting instance.
[0116] The accuracy of the calculation of the mean values over the frequency subbands may vary with changes to channel conditions. Better estimation accuracies may be associated with combining values of multiple intermediate KPI value (e.g., combining estimates from all frequency sub-bands, all prediction time instances, and / or all layers) to form one single statistical value, which is referred to herein as the combined mean value. The combined mean value may be reported to the network access node.
[0117] As the reliability of each intermediate KPI value will likely vary over the different domains (as illustrated during Figure 6, which illustrates a change in CSI accuracy over time for each of a plurality of different spatial domains (e.g., layers 1 to 4)), the combination of mean values may be formed by applying a weighted sum over at least one of frequency, time, and spatial (e.g., layer) domains, where the weights are calculated from the expected reliability of each contribution. For example, a statistical mean of, and variance of, the intermediate KPI values may be determined per domain,with the weights for each domain being applied relative to an estimated reliability per domain and monitoring instance.
[0118] The calculation of the weighted sum may be configured for different numbers of subdomains within each domain (e.g., for different numbers of sub-bands Nf, prediction time instances Tpand number of layers Lm). Configuration parameters that define the number of sub-bands Nf, number of prediction time instances TPand / or number of layers Lm may be defined (e.g., derived and / or calculated) at the UE and reported to the network access node, or may be configured at the UE by a network access node.
[0119] Further, when a weighted sum is reported, it may be useful for the report defined by the format to comprise reliability values for each domain and / or for each reporting instance. To this effect, the UE may use long term observations for defining a value to be used as the UE’s reference intermediate KPI value predictor and / or the ML intermediate KPI value predictor that provides a relative mean value and variance values. The reference intermediate KPI value(s) may be used to learn the estimated performance variation over each of the domains, which may be useful for scenarios in which the CSI prediction performance improves and degrades non-linearly.
[0120] Reporting of multiple layers with a weighted sum applied either at the UE or at the network access node after reporting of reliability information may ensure that all available prediction performance estimates are used to get the most accurate inference result.
[0121] A codebook may be used for efficiently reporting the weighted sum example, depending on the information that is to be reported by the UE. For example, the UE may apply the weighted sum operation over all domains and report a single weighted mean key performance indicator value to the network access node, or the UE may report the mean values for each domain and the network access node may apply the weights per domain based on the reported relative weights.
[0122] In this latter case, the format for the report may cause the UE to report the applied weight values in addition to the mean intermediate KPI value(s). This may be performed by reporting the relative weights per domain as an identifier of a codebook. This is illustrated in the example of Figure 7, where the different codebook identifiers (IDs) CB1 to CB6 indicate different relative weights per domain to to tN. codebook with different relative weight per domain to to tN. The UE may select a best fitting codebookidentifier from a set of codebook identifiers comprised in a predefined codebook the best fitting codebook ID to indicate the relative weights to the network access node.
[0123] A similar codebook method may be used for reporting the mean intermediate KPI values per domain as difference values related a strongest mean value. This mean value difference codebook method may allow the network access node to combine the mean values as applicable.
[0124] The multi-instance report over N monitoring instances with inherent error correction may combine extremely low reporting overhead (e.g., only two or even only 1 bit per monitoring instance report) with an increasing level of accuracy with an increasing number of monitoring reports. The very low reporting overhead may allow the UE to maintain a constant background performance monitoring mode, where the UE reports, for each CSI prediction, one to two extra bits related to the monitoring performance.
[0125] Reporting of relative SGCS values between a reference predictor (e.g., the above-mentioned zero order hold) and the ML model (e.g., AI / ML) predictor may be useful for reporting as it is expected that there will only be small deviations between these values (e.g., in the difference between these values, labelled DSGCS value herein). Generally, the absolute SGCS values, e.g., of the reference predictor may not be reported. Nonetheless, a network access node might benefit from knowing at least the rough value range of the absolute SGCS values as for higher SGCS values a certain DSGCS value might provide more performance gain. The format to be used for reporting intermediate KPI values by the UE may therefore be changed to occasionally include a coarse absolute SGCS value.
[0126] The multi-instance report over N monitoring instances with inherent error correction can be very efficient for high numbers N greater than 10 or higher. Typically, the number of reports may be configured by the network access node in advance. However, as the UE reports the mean values including correction of previous reports constantly, the reporting may be performed such that there are some reports without a predefined value of N. Alternatively, the network access node may collect a number of reports from the UE, and determine the number, N, based on the collected reports.
[0127] Assuming the reporting with inherent error correction uses per reporting instance a 2-bit quantization then the maximum estimation error per instance would be e < ±0.125, which can be reduced for N equal to 10 to ew < ±0.0125. Even a onebit quantization is possible, where e1< ±0.25, which can be reduced for N equal to 10 to e1io < ±0.025.
[0128] The reporting of mean values in combination with reliability information allows a network access node to combine multiple reports as a weighted sum taking the reliability of each report into account.
[0129] In case of very low reliability the UE might not report values relating to an intermediate KPI value at all. To this effect, the network access node may configure, at the UE, a minimum reliability threshold THreithat is to be met before the UE may report in accordance with the indicated format.
[0130] The reliability of the reference predictor and that of the AI / ML predictor might vary in a different way. Therefore, it might be useful to report different reliability values Rref and Rpredict for the reference predictor as well as the AI / ML predictor and to define different thresholds THrefand THpredict.
[0131] The reliability of certain SGCS value estimates might depend on various parameters like the mean value itself (e.g., small values will be typically less reliable), the variance (larger variance indicates a lower reliability), SINR, channel delay spread and others. It might be difficult to identify all relevant aspects defining the reliability and, in addition, to define the relative reliabilities and / or the optimum relative weights from the reliability values. Therefore, an AI / ML model inferring the reliability and respectively the optimal weights used for the calculation of the weighted sum is a natural choice.
[0132] The following illustrates how one of more of the above-mentioned techniques may be employed in combination with each other, with reference to Figures 8A and 8B.
[0133] Generally, the SGCS^tvalues over different layers I and different prediction time instances n4have different reliabilities so that the averaging over all layers and prediction instances might even degrade the accuracy for the performance estimation. Figure 8A illustrates for example, a significantly varying histograms for one UE at the different time instances n4= 1, 2, and 3 for the non-predicted (e.g., ZOH value) as well as the predicted SGCS values (e.g., long short term memory (LSTM) values). The histogram corresponding to the first time instance is labelled as 801, the histogram corresponding to the second time instance is labelled as 802, and the histogram corresponding to the third time instance is labelled as 803. One option is that the UEreports multiple median and variance SGCS values over the frequency domain individually for each relevant layer and relevant prediction time instance. This might lead to a relatively high reporting overhead. Therefore, the domain specific reliabilities rin4may be calculated per layer I and per prediction time instance n4and to calculate the weighted sum over all layers I and prediction time instance n4to get the weighted SGCSi value for the i - th inference reporting instance:SGCSt= — Xn3’?rm. (— Zn3-i SGCS"3}
[0134] It is therefore proposed that when the UE calculates the SGCSi value of one inference reporting instance over multiple layers I and multiple prediction time instances n4, then the UE may also report the weighted sum over all layers and prediction time instances.
[0135] Figure 8B illustrates four bar charts illustrating SGCS values with median and variance for 15 different example UEs collected over 10 time instances for the AI / ML predictor on the left and the non-predicted ZOH reference predictor at the right. The top two bar charts illustrate example SGCS values (with their associated mean and variance a first prediction time instance with tpredict = 5 ms. The bottom two bar charts in Figure 8B illustrate example SGCS values (with their associated mean and variance a fourth prediction time instance with tpredict = 20 ms.
[0136] In Figure 8B the median and variance values per UE have been calculated over I = 10 inference reporting instances to ensure a sufficient statistic for the calculation of the SGCS values. This can be achieved, for example, by averaging the SGCSi value reports of I = 10 inference reporting instances. Independent of the content of one SGCSi value, (e.g., either the report for a single layer Z and single prediction time instance n4) the reported values have may be quantized.
[0137] Assuming I » 1 then a reporting scheme with a low number of bits per report and simultaneously a low quantization error e is possible. This can be achieved by adding to each SGCSi value first the quantization error of the previous reportso that we have SGCS'i=SGCSi - e^. SGCS'i is then quantized again by the quantization function Q(. to generate the next report i. After the gNB has received the last report I it can average over all reports, i.e., divide by I the sum SGCS = i^SGCS'i to get the estimated value SGCS. Assume a 2-bit quantization function Q(.) The quantizationerror etper monitoring report i will be et= ±0.125 while after I = 10 reports the error can be reduced to e / =10= ±0.0125.
[0138] It may therefore be proposed to report SGCS values over multiple reporting instances I » 1 with inherent error correctionso that a low bit quantization may be simultaneously achieved in combination with a very low estimation error e / =10.
[0139] Figures 9 and 10 illustrate one or more methods that may be employed by an apparatus as described herein. The one or more methods may find correspondence with one or more of the above-mentioned implementation examples.
[0140] Figure 9 illustrates one or more methods that may be performed by an apparatus. The apparatus may comprise (or be comprised in) a user equipment, such as that described with reference to any of Figures 1 and 3.
[0141] During 901, the apparatus deploys, at the apparatus, a machine learning model for estimating one or more characteristics of a communication channel. The one or more characteristics of a communication channel may be estimated, for example, based on one or more measurements performed by the apparatus on a reference signal, such as a channel state information reference signal.
[0142] During 902, the apparatus transmits to a network access node, capability information, wherein the capability information indicates one or more formats that may be used by the apparatus for transmitting a report comprising information indicating one or more intermediate key performance indicator values of the machine learning model. The one or more formats may comprise a first format. It is understood that 902 may be performed before 901. The network access node may comprise (or be comprised in) the apparatus of Figure 10.
[0143] During 903, the apparatus receives, from the network access node, a configuration for reporting the information indicating one or more intermediate key performance indicator values of the machine learning model.
[0144] During 904, the apparatus selects a first format from the one or more formats for reporting the information indicating the one or more intermediate key performance indicator values to the network access node based on the received configuration.
[0145] During 905, the apparatus transmits the report comprising the information indicating the one or more intermediate key performance indicator values to the network access node in accordance with the first format.
[0146] The information indicating the one or more intermediate key performance indicator values may comprise an absolute value of the one or more intermediate key performance indicator values. The information indicating the one or more intermediate key performance indicator values may comprise information indicating a difference between an intermediate KPI value of the machine learning model and a reference intermediate KPI value (e.g., the immediately preceding KPI value when a zero order hold is used). The information indicating the one or more intermediate key performance indicator values may comprise statistical information relating to the one or more intermediate KPI values, such as a mean intermediate KPI value. The information indicating the one or more intermediate key performance indicator values may relate to a single monitoring period, or to multiple monitoring periods.
[0147] The transmitting the report comprising the information indicating the one or more intermediate key performance indicator values to the network access node in accordance with the first format may comprise at least one of: transmitting a report comprising information indicating the one or more intermediate key performance indicator values per frequency sub-band, per layer, or per prediction time instance of the machine learning model.
[0148] Figure 10 illustrates one or more methods that may be performed by an apparatus. The apparatus may comprise (or be comprised in) a network access node, such as that described with reference to any of Figures 1 and 2.
[0149] During 1001, the apparatus receives, from a user equipment, capability information, wherein the capability information indicates one or more formats that may be used by the user equipment for transmitting a report comprising information indicating one or more intermediate key performance indicator values for estimating one or more characteristics of a communication channel of a machine learning model deployed at the user equipment, wherein the one or more formats comprises a first format. The user equipment may comprise, or be comprised in, the apparatus of Figure 9.
[0150] During 1002, the apparatus determines, based on the capability information, a configuration for reporting the information indicating the one or more intermediate key performance indicator values.
[0151] During 1003, the apparatus transmits, to the user equipment, information indicating the configuration.
[0152] During 1004, the apparatus receives, from the user equipment, the report comprising the information indicating the one or more intermediate key performance indicator values in a first format, the first format being one of the one or more formats.
[0153] The receiving the report comprising the information indicating the one or more intermediate key performance indicator values in the first format may comprise at least one of: receiving a report comprising information indicating the one or more intermediate key performance indicator values per frequency sub-band, per layer, or per prediction time instance of the machine learning model.
[0154] The following may apply in respect of any (e.g., one or more, including all) of the methods of Figures 9 and 10.
[0155] The intermediate key performance indicator values may comprise at least one of: a squared generalized cosine similarity value, or a value representing a plurality of squared generalized cosine similarity values.
[0156] The information indicating the one or more intermediate key performance indicator values may indicate, for each of one or more determined intermediate key performance indicator value, a difference between that determined intermediate key performance indicator value and a respective reference intermediate key performance indicator value, and the configuration defines a range of said differences that are to be reported via the first format.
[0157] The first format may comprise a plurality of bits, each of said bits corresponding to at least one of: a first subrange of values of said range, a second subrange of values of said range, a first out-of-range indication for a difference value that is above the upper limit of the range, or a second out-of-range indication for a difference value that is below the lower limit of the range.
[0158] The first subrange may define a first subrange of values, and the second subrange may define a second subrange of values. The first and the second subrange of values may be equally sized. The first and second subrange of values may be unequally sized.
[0159] The reference intermediate key performance indicator value may be based on a last channel observation obtained relative to a ground truth determined at a prediction time instance
[0160] The one or more intermediate key performance indicator values may comprise a value representing a plurality of squared generalized cosine similarity values by comprising at least one of: a mean or median value of a plurality of determinedintermediate key performance indicator values or a mean or median value of, for each of the plurality of determined intermediate key performance indicator values, a difference between that determined intermediate key performance indicator value and a respective reference intermediate key performance indicator value.
[0161] The report may further comprise information indicating a reliability of the one or more intermediate key performance indicator values.
[0162] The information indicating the reliability may comprise at least one of: a variance of the one or more intermediate key performance indicator values, a signal to interference and noise ratio, ora delay spread. The delay spread may be considered to be a measure of how far the delays of the multipath components comprising a radio channel vary statistically.
[0163] The report may comprise information indicating one or more intermediate key performance indicator values for a plurality of at least one of the following domains: a frequency domain, a spatial domain, or a time domain.
[0164] The one or more intermediate key performance indicator values comprised in the report may be, for each of the one or more intermediate key performance indictor values, weighted by a reliability of that intermediate key performance indicator value.
[0165] The report may comprise an identifier of a codebook value, wherein the identifier of the codebook value indicates the one or more intermediate key performance indictor values across at least one of the frequency domain, spatial domain, and time domain.
[0166] The identifier of the codebook value may indicate the one or more intermediate key performance indictor values across a plurality of the frequency domain, spatial domain, and time domain.
[0167] The identifier of the codebook may identify a respective relative value of at least one of the frequency domain, spatial domain, or time domain, wherein a relative value indicates a difference between the value being reported and a corresponding strongest intermediate key performance indictor value.
[0168] The one or more intermediate key performance indicator values may comprise a multi-instance report. For example, the one or more intermediate key performance indicator values may comprise multiple statistical values, each of the multiple statistical values being associated with a respective error correction value, and each of the multiple statistical values corresponding to a respective monitoring instance.
[0169] The apparatus may receive another report from the user equipment in accordance with the first format, the another report comprising respective information indicating one or more corresponding intermediate key performance indicator values, wherein the one or more corresponding intermediate key performance indicator values comprises multiple corresponding statistical values, each of the multiple corresponding statistical values being associated with a respective error correction value that is based on respective error correction value(s) comprised in the report, and perform an averaging operation across at least values comprised in the report and the another report to obtain a mean key performance indicator value.
[0170] The apparatus may transmit, to the network access node, another report in accordance with the first format, the another report comprising respective information indicating one or more corresponding intermediate key performance indicator values, wherein the one or more corresponding intermediate key performance indicator values comprises multiple corresponding statistical values, each of the multiple corresponding statistical values being associated with a respective error correction value that is based on respective error correction value(s) comprised in the report.
[0171] It should be understood that the apparatuses may comprise or be coupled to other units or modules etc., such as radio parts or radio heads, used in or for transmission and / or reception. Although the apparatuses have been described as one entity, different modules and memory may be implemented in one or more physical or logical entities.
[0172] It is noted that whilst some embodiments have been described in relation to 5G networks, similar principles can be applied in relation to other networks and communication systems. Therefore, although certain embodiments were described above by way of example with reference to certain example architectures for wireless networks, technologies and standards, embodiments may be applied to any other suitable forms of communication systems than those illustrated and described herein.
[0173] It is also noted herein that while the above describes example embodiments, there are several variations and modifications which may be made to the disclosed solution without departing from the scope of the present invention.
[0174] As used herein, “at least one of the following: ” and “at least one of ” and similar wording, where the list of two or more elements are joined by “and” or “or”, mean at least any one of the elements, or at least any two or more of the elements, or at least all the elements.
[0175] In general, the various embodiments may be implemented in hardware or special purpose circuitry, software, logic or any combination thereof. Some aspects of the disclosure may be implemented in hardware, while other aspects may be implemented in firmware or software which may be executed by a controller, microprocessor or other computing device, although the disclosure is not limited thereto. While various aspects of the disclosure may be illustrated and described as block diagrams, flow charts, or using some other pictorial representation, it is well understood that these blocks, apparatus, systems, techniques or methods described herein may be implemented in, as non-limiting examples, hardware, software, firmware, special purpose circuits or logic, general purpose hardware or controller or other computing devices, or some combination thereof.
[0176] As used herein, the term “circuitry” may refer to one or more or all of the following:(a) hardware-only circuit implementations (such as implementations in only analog and / or digital circuitry) and(b) combinations of hardware circuits and software, such as (as applicable):(i) a combination of analog and / or digital hardware circuit(s) with software / firmware and(ii) any portions of hardware processor(s) with software (including digital signal processor(s)), software, and memory(ies) that work together to cause an apparatus, such as a mobile phone or server, to perform various functions) and(c) hardware circuit(s) and or processor(s), such as a microprocessor(s) or a portion of a microprocessor(s), that requires software (e.g., firmware) for operation, but the software may not be present when it is not needed for operation.”
[0177] This definition of circuitry applies to all uses of this term herein, including in any claims. As a further example, as used herein, the term circuitry also covers an implementation of merely a hardware circuit or processor (or multiple processors) or portion of a hardware circuit or processor and its (or their) accompanying software and / or firmware. The term circuitry also covers, for example and if applicable to the particular claim element, a baseband integrated circuit or processor integrated circuit for a mobile device or a similar integrated circuit in server, a cellular network device, or other computing or network device.
[0178] The embodiments of this disclosure may be implemented by computer software executable by a data processor of the mobile device, such as in the processor entity,or by hardware, or by a combination of software and hardware. Computer software or program, also called program product, including software routines, applets and / or macros, may be stored in any apparatus-readable data storage medium and they comprise program instructions to perform particular tasks. A computer program product may comprise one or more computer-executable components which, when the program is run, are configured to carry out embodiments. The one or more computer-executable components may be at least one software code or portions of it.
[0179] Further in this regard it should be noted that any blocks of the logic flow as in the Figures may represent program steps, or interconnected logic circuits, blocks and functions, or a combination of program steps and logic circuits, blocks and functions. The software may be stored on such physical media as memory chips, or memory blocks implemented within the processor, magnetic media such as hard disk or floppy disks, and optical media such as for example DVD and the data variants thereof, CD. The physical media is a non-transitory media.
[0180] The term “non-transitory,” as used herein, is a limitation of the medium itself (i.e., tangible, not a signal) as opposed to a limitation on data storage persistency (e.g., RAM vs. ROM).
[0181] The memory may be of any type suitable to the local technical environment and may be implemented using any suitable data storage technology, such as semiconductor-based memory devices, magnetic memory devices and systems, optical memory devices and systems, fixed memory and removable memory. The data processors may be of any type suitable to the local technical environment, and may comprise one or more of general purpose computers, special purpose computers, microprocessors, digital signal processors (DSPs), application specific integrated circuits (ASIC), FPGA, gate level circuits and processors based on multi core processor architecture, as non-limiting examples.
[0182] Various example embodiments of the disclosure may be practiced in various components such as integrated circuit modules. The design of integrated circuits is by and large a highly automated process. Complex and powerful software tools are available for converting a logic level design into a semiconductor circuit design ready to be etched and formed on a semiconductor substrate.
[0183] The scope of protection sought for various example embodiments of the disclosure is set out by the independent claims. The example embodiments and features thereof, if any, described in this disclosure that do not fall under the scope ofthe independent claims are to be interpreted as examples useful for understanding various example embodiments of the disclosure.
[0184] The foregoing description has provided, by way of non-limiting and illustrative examples, a full and informative description of the various example embodiments of this disclosure. However, various modifications and adaptations may become apparent to those skilled in the relevant arts in view of the foregoing description, when read in conjunction with the accompanying drawings and the claims. However, all such and similar modifications of the teachings will still fall within the various example embodiments of the disclosure as set forth in the claims. By way of non-limiting and illustrative example, there is a further example embodiment comprising a combination of one or more example embodiments with any of the other example embodiments previously discussed.
Claims
CLAIMS1) An apparatus comprising:at least one processor; andat least one memory comprising code that, when executed by the at least one processor, causes the apparatus to perform:deploying, at the apparatus, a machine learning model for estimating one or more characteristics of a communication channel;transmitting, to a network access node, capability information, wherein the capability information indicates one or more formats that may be used by the apparatus for transmitting a report comprising information indicating one or more intermediate key performance indicator values of the machine learning model;receiving, from the network access node, a configuration for reporting the information indicating one or more intermediate key performance indicator values of the machine learning model;selecting a first format from the one or more formats for reporting the information indicating the one or more intermediate key performance indicator values to the network access node based on the received configuration; and transmitting the report comprising the information indicating the one or more intermediate key performance indicator values to the network access node in accordance with the first format.2) An apparatus as claimed in claim 1, wherein the transmitting the report comprising the information indicating the one or more intermediate key performance indicator values to the network access node in accordance with the first format comprises at least one of: transmitting a report comprising information indicating the one or more intermediate key performance indicator values per frequency sub-band, per layer, or per prediction time instance of the machine learning model.3) An apparatus comprising:at least one processor; andat least one memory comprising code that, when executed by the at least one processor, causes the apparatus to perform:receiving, from a user equipment, capability information, wherein the capability information indicates one or more formats that may be used by the user equipment for transmitting a report comprising information indicating one or more intermediate key performance indicator values for estimating one or more characteristics of a communication channel of a machine learning model deployed at the user equipment ;determining, based on the capability information, a configuration for reporting the information indicating the one or more intermediate key performance indicator values;transmitting, to the user equipment, information indicating the configuration; andreceiving, from the user equipment, the report comprising the information indicating the one or more intermediate key performance indicator values in a first format, the first format being one of the one or more formats.4) An apparatus as claimed in claim 3, wherein the receiving the report comprising the information indicating the one or more intermediate key performance indicator values in the first format comprises at least one of: receiving a report comprising information indicating the one or more intermediate key performance indicator values per frequency sub-band, per layer, or per prediction time instance of the machine learning model.5) An apparatus as claimed in any preceding claim, wherein the one or more intermediate key performance indicator values comprises at least one of: a squared generalized cosine similarity value, or a value representing a plurality of squared generalized cosine similarity values.6) An apparatus as claimed in any preceding claim, wherein the information indicating the one or more intermediate key performance indicator values indicates, for each of one or more determined intermediate key performance indicator value, a difference between that determined intermediate key performance indicator value and a respective reference intermediate key35performance indicator value, and the configuration defines a range of said differences that are to be reported via the first format.7) An apparatus as claimed in any of claims 1 to 4, wherein the first format comprises a plurality of bits, each of said bits corresponding to at least one of: a first subrange of values of said range, a second subrange of values of said range, a first out-of-range indication for a difference value that is above the upper limit of the range, or a second out-of-range indication for a difference value that is below the lower limit of the range.8) An apparatus as claimed in claim 7, wherein the first subrange defines a first subrange of values, and the second subrange defines a second subrange of values, and wherein the first and second subrange of values are not equally sized.9) An apparatus as claimed in any of claims 6 to 8, wherein the reference intermediate key performance indicator value is based on a last channel observation obtained relative to a ground truth determined at a prediction time instance10)An apparatus as claimed in any of claims 5 to 9, wherein the one or more intermediate key performance indicator values comprises the value representing the plurality of squared generalized cosine similarity values by comprising at least one of: a mean or median value of a plurality of determined intermediate key performance indicator values or a mean or median value of, for each of the plurality of determined intermediate key performance indicator values, a difference between that determined intermediate key performance indicator value and a respective reference intermediate key performance indicator value.11)An apparatus as claimed in any of claims 5 to 10, wherein the report further comprises information indicating a reliability of the one or more intermediate key performance indicator values.)An apparatus as claimed in claim 11, wherein the information indicating the reliability comprises at least one of: a variance of the one or more intermediate key performance indicator values, a signal to interference and noise ratio, or a delay spread.)An apparatus as claimed in any preceding claim, wherein the report comprises information indicating one or more intermediate key performance indicator values for a plurality of at least one of the following domains: a frequency domain, a spatial domain, or a time domain.)An apparatus as claimed in claim 13, wherein the one or more intermediate key performance indicator values comprised in the report is, for each of the one or more intermediate key performance indictor values, weighted by a reliability of that intermediate key performance indicator value.)An apparatus as claimed in any preceding claim, wherein the report comprises an identifier of a codebook value, wherein the identifier of the codebook value indicates the one or more intermediate key performance indictor values across at least one of the frequency domain, spatial domain, and time domain.)An apparatus as claimed in claim 15, wherein the identifier of the codebook value indicates the one or more intermediate key performance indictor values across a plurality of the frequency domain, spatial domain, and time domain.)An apparatus as claimed in any of claims 15 to 16, wherein the identifier of the codebook identifies a respective relative value of at least one of the frequency domain, spatial domain, or time domain, wherein a relative value indicates a difference between the value being reported and a corresponding strongest intermediate key performance indictor value.)An apparatus as claimed in any preceding claim, wherein the one or more intermediate key performance indicator values comprises multiple statistical values, each of the multiple statistical values being associated with a respectiveerror correction value, and each of the multiple statistical values corresponding to a respective monitoring instance.)An apparatus as claimed in claim 18 when dependent on any of claims 3 to 4, wherein the apparatus is further caused to perform:receiving another report from the user equipment in accordance with the first format, the another report comprising respective information indicating one or more corresponding intermediate key performance indicator values, wherein the one or more corresponding intermediate key performance indicator values comprises multiple corresponding statistical values, each of the multiple corresponding statistical values being associated with a respective error correction value that is based on respective error correction value(s) comprised in the report; andperforming an averaging operation across at least values comprised in the report and the another report to obtain a mean key performance indicator value.)An apparatus as claimed in claim 18 when dependent on any of claims 1 to 2, wherein the apparatus is further caused to perform:transmitting, to the network access node, another report in accordance with the first format, the another report comprising respective information indicating one or more corresponding intermediate key performance indicator values, wherein the one or more corresponding intermediate key performance indicator values comprises multiple corresponding statistical values, each of the multiple corresponding statistical values being associated with a respective error correction value that is based on respective error correction value(s) comprised in the report.)An apparatus comprising means for performing:deploying, at the apparatus, a machine learning model for estimating one or more characteristics of a communication channel;transmitting, to a network access node, capability information, wherein the capability information indicates one or more formats that may be used by the apparatus for transmitting a report comprising information indicating one or38more intermediate key performance indicator values of the machine learning model;receiving, from the network access node, a configuration for reporting the information indicating one or more intermediate key performance indicator values of the machine learning model;selecting a first format from the one or more formats for reporting the information indicating the one or more intermediate key performance indicator values to the network access node based on the received configuration; and transmitting the report comprising the information indicating the one or more intermediate key performance indicator values to the network access node in accordance with the first format.)An apparatus as claimed in claim 21, wherein the transmitting the report comprising the information indicating the one or more intermediate key performance indicator values to the network access node in accordance with the first format comprises at least one of: transmitting a report comprising information indicating the one or more intermediate key performance indicator values per frequency sub-band, per layer, or per prediction time instance of the machine learning model.)An apparatus comprising means for performing:receiving, from a user equipment, capability information, wherein the capability information indicates one or more formats that may be used by the user equipment for transmitting a report comprising information indicating one or more intermediate key performance indicator values for estimating one or more characteristics of a communication channel of a machine learning model deployed at the user equipment ;determining, based on the capability information, a configuration for reporting the information indicating the one or more intermediate key performance indicator values;transmitting, to the user equipment, information indicating the configuration; and39receiving, from the user equipment, the report comprising the information indicating the one or more intermediate key performance indicator values in a first format, the first format being one of the one or more formats.)An apparatus as claimed in claim 23, wherein the receiving the report comprising the information indicating the one or more intermediate key performance indicator values in the first format comprises at least one of: receiving a report comprising information indicating the one or more intermediate key performance indicator values per frequency sub-band, per layer, or per prediction time instance of the machine learning model.)An apparatus as claimed in any of claims 21 to 24, wherein the one or more intermediate key performance indicator values comprises at least one of: a squared generalized cosine similarity value, or a value representing a plurality of squared generalized cosine similarity values.)An apparatus as claimed in any of claims 21 to 25, wherein the information indicating the one or more intermediate key performance indicator values indicates, for each of one or more determined intermediate key performance indicator value, a difference between that determined intermediate key performance indicator value and a respective reference intermediate key performance indicator value, and the configuration defines a range of said differences that are to be reported via the first format.)An apparatus as claimed in any of claims 21 to 24, wherein the first format comprises a plurality of bits, each of said bits corresponding to at least one of: a first subrange of values of said range, a second subrange of values of said range, a first out-of-range indication for a difference value that is above the upper limit of the range, or a second out-of-range indication for a difference value that is below the lower limit of the range.)An apparatus as claimed in claim 27, wherein the first subrange defines a first subrange of values, and the second subrange defines a second subrange of40values, and wherein the first and second subrange of values are not equally sized.)An apparatus as claimed in any of claims 26 to 28, wherein the reference intermediate key performance indicator value is based on a last channel observation obtained relative to a ground truth determined at a prediction time instance)An apparatus as claimed in any of claims 25 to 29, wherein the one or more intermediate key performance indicator values comprises the value representing the plurality of squared generalized cosine similarity values by comprising at least one of: a mean or median value of a plurality of determined intermediate key performance indicator values or a mean or median value of, for each of the plurality of determined intermediate key performance indicator values, a difference between that determined intermediate key performance indicator value and a respective reference intermediate key performance indicator value.)An apparatus as claimed in any of claims 25 to 30, wherein the report further comprises information indicating a reliability of the one or more intermediate key performance indicator values.)An apparatus as claimed in claim 31, wherein the information indicating the reliability comprises at least one of: a variance of the one or more intermediate key performance indicator values, a signal to interference and noise ratio, or a delay spread.)An apparatus as claimed in any preceding claim, wherein the report comprises information indicating one or more intermediate key performance indicator values for a plurality of at least one of the following domains: a frequency domain, a spatial domain, or a time domain.)An apparatus as claimed in claim 33, wherein the one or more intermediate key performance indicator values comprised in the report is, for each of the one or41more intermediate key performance indictor values, weighted by a reliability of that intermediate key performance indicator value.)An apparatus as claimed in any of claims 21 to 24, wherein the report comprises an identifier of a codebook value, wherein the identifier of the codebook value indicates the one or more intermediate key performance indictor values across at least one of the frequency domain, spatial domain, and time domain.)An apparatus as claimed in claim 35, wherein the identifier of the codebook value indicates the one or more intermediate key performance indictor values across a plurality of the frequency domain, spatial domain, and time domain.)An apparatus as claimed in any of claims 35 to 36, wherein the identifier of the codebook identifies a respective relative value of at least one of the frequency domain, spatial domain, or time domain, wherein a relative value indicates a difference between the value being reported and a corresponding strongest intermediate key performance indictor value.)An apparatus as claimed in any of claims 21 to 37, wherein the one or more intermediate key performance indicator values comprises multiple statistical values, each of the multiple statistical values being associated with a respective error correction value, and each of the multiple statistical values corresponding to a respective monitoring instance.)An apparatus as claimed in claim 38 when dependent on any of claims 23 to 24, wherein the apparatus further comprises means for performing:receiving another report from the user equipment in accordance with the first format, the another report comprising respective information indicating one or more corresponding intermediate key performance indicator values, wherein the one or more corresponding intermediate key performance indicator values comprises multiple corresponding statistical values, each of the multiple corresponding statistical values being associated with a respective error correction value that is based on respective error correction value(s) comprised in the report; andperforming an averaging operation across at least values comprised in the report and the another report to obtain a mean key performance indicator value.)An apparatus as claimed in claim 28 when dependent on any of claims 21 to 22, wherein the apparatus further comprises means for performing:transmitting, to the network access node, another report in accordance with the first format, the another report comprising respective information indicating one or more corresponding intermediate key performance indicator values, wherein the one or more corresponding intermediate key performance indicator values comprises multiple corresponding statistical values, each of the multiple corresponding statistical values being associated with a respective error correction value that is based on respective error correction value(s) comprised in the report.)A method for an apparatus, the method comprising:deploying, at the apparatus, a machine learning model for estimating one or more characteristics of a communication channel;transmitting, to a network access node, capability information, wherein the capability information indicates one or more formats that may be used by the apparatus for transmitting a report comprising information indicating one or more intermediate key performance indicator values of the machine learning model;receiving, from the network access node, a configuration for reporting the information indicating one or more intermediate key performance indicator values of the machine learning model;selecting a first format from the one or more formats for reporting the information indicating the one or more intermediate key performance indicator values to the network access node based on the received configuration; and transmitting the report comprising the information indicating the one or more intermediate key performance indicator values to the network access node in accordance with the first format.)A method for an apparatus, the method comprising:43receiving, from a user equipment, capability information, wherein the capability information indicates one or more formats that may be used by the user equipment for transmitting a report comprising information indicating one or more intermediate key performance indicator values for estimating one or more characteristics of a communication channel of a machine learning model deployed at the user equipment;determining, based on the capability information, a configuration for reporting the information indicating the one or more intermediate key performance indicator values;transmitting, to the user equipment, information indicating the configuration; andreceiving, from the user equipment, the report comprising the information indicating the one or more intermediate key performance indicator values in a first format, the first format being one of the one or more formats.)A computer program comprising instructions which, when the program is executed by a computer of an apparatus, cause the apparatus to carry out: deploying, at the apparatus, a machine learning model for estimating one or more characteristics of a communication channel;transmitting, to a network access node, capability information, wherein the capability information indicates one or more formats that may be used by the apparatus for transmitting a report comprising information indicating one or more intermediate key performance indicator values of the machine learning model;receiving, from the network access node, a configuration for reporting the information indicating one or more intermediate key performance indicator values of the machine learning model;selecting a first format from the one or more formats for reporting the information indicating the one or more intermediate key performance indicator values to the network access node based on the received configuration; and transmitting the report comprising the information indicating the one or more intermediate key performance indicator values to the network access node in accordance with the first format.44)A computer program comprising instructions which, when the program is executed by a computer of an apparatus, cause the apparatus to carry out: receiving, from a user equipment, capability information, wherein the capability information indicates one or more formats that may be used by the user equipment for transmitting a report comprising information indicating one or more intermediate key performance indicator values for estimating one or more characteristics of a communication channel of a machine learning model deployed at the user equipment;determining, based on the capability information, a configuration for reporting the information indicating the one or more intermediate key performance indicator values;transmitting, to the user equipment, information indicating the configuration; andreceiving, from the user equipment, the report comprising the information indicating the one or more intermediate key performance indicator values in a first format, the first format being one of the one or more formats.45
