Delay aware life cycle management operations for beam prediction

Abstract

Delay aware life cycle management operations for beam prediction may be provided. A method of delay aware life cycle management operations for beam prediction may include receiving, from a network entity, a configuration for artificial intelligence / machine learning (AI / ML) beam management. The configuration may indicate a reporting frequency for AI / ML beam management as aperiodic reporting. The method may also include determining one or more delay requirements associated with the AI / ML beam management, determining a time instance based on the determined one or more delay requirements associated with the AI / ML beam management, performing the AI / ML beam management, and transmitting, via aperiodic reporting to the network entity during the time instance, a report indicating one or more inference results of the AI / ML beam management.

Description

TITLE: DELAY AWARE LIFE CYCLE MANAGEMENT OPERATIONS FOR BEAM PREDICTION TECHNICAL FIELD:

[0001] Some exemplary embodiments may generally relate to mobile or wireless telecommunication systems, such as Long Term Evolution (LTE) or fifth generation (5G) or sixth generation (6G) new radio (NR) access technology, or other communications systems. For example, certain exemplary embodiments may relate to implementing delay aware life cycle management operations for beam prediction. BACKGROUND:

[0002] Examples of mobile or wireless telecommunication systems may include the Universal Mobile Telecommunications System (UMTS) Terrestrial Radio Access Network (UTRAN), Long Term Evolution (LTE) Evolved UTRAN (E-UTRAN), LTE-Advanced (LTE-A), MulteFire, LTE-A Pro, fifth generation (5G) radio access technology or new radio (NR) access technology, and / or sixth generation (6G) radio access technology.5G and 6G wireless systems refer to the next generation (NG) of radio systems and network architecture.5G and 6G network technology are mostly based on new radio (NR) technology, but the 5G (or NG) network can also build on E-UTRAN radio. It is estimated that NR may provide bitrates on the order of 10-20 Gbit / s or higher and may support at least enhanced mobile broadband (eMBB) and ultra-reliable low- latency communication (URLLC) as well as massive machine-type communication (mMTC). NR is expected to deliver extreme broadband and ultra-robust, low-latency connectivity and massive networking to support the Internet of Things (IoT). SUMMARY:

[0003] Various exemplary embodiments may provide an apparatus comprising at least one processor and at least one memory storing instructions that, when executed by the at least one processor, cause the apparatus at least to receive, from a network entity, a configuration for artificial intelligence / machine learning (AI / ML) beam management. The configuration may indicate a reporting frequency for the AI / ML beam management as periodic reporting. The apparatus may also be caused to determine one or more delay requirements associated with the AI / ML beam management, determine a time instance based on the determined one or more delay requirements associated with the AI / ML beam management, perform the AI / ML beam management, and transmit, via periodic reporting to the network entity during the time instance, a report indicating one or more inference results of the AI / ML beam management.

[0004] Certain exemplary embodiments may provide an apparatus comprising at least one processor and at least one memory storing instructions that, when executed by the at least one processor, cause the apparatus at least to determine a configuration for artificial intelligence / machine learning (AI / ML) beammanagement to be performed by a user device and determine one or more delay requirements associated with the AI / ML beam management. The apparatus may also be caused to determine a time instance based on the determined one or more delay requirements associated with the AI / ML beam management and transmit to the user device, the configuration for AI / ML beam management. The configuration may indicate a reporting frequency for the AI / ML beam management as periodic reporting. The apparatus may further be caused to receive, via periodic reporting from the user device during the time instance, a report indicating one or more inference results of the AI / ML beam management.

[0005] Some exemplary embodiments may provide an apparatus comprising at least one processor and at least one memory storing instructions that, when executed by the at least one processor, cause the apparatus at least to receive, from a network entity, a configuration for artificial intelligence / machine learning (AI / ML) beam management. The configuration may indicate a reporting frequency for the AI / ML beam management as semi-persistent reporting. The apparatus may also be caused to determine one or more delay requirements associated with the AI / ML beam management, determine a time instance based on the determined one or more delay requirements associated with the AI / ML beam management, perform the AI / ML beam management, and transmit, via semi-persistent reporting to the network entity during the time instance, a report indicating one or more inference results of the AI / ML beam management.

[0006] Certain exemplary embodiments may provide an apparatus comprising at least one processor and at least one memory storing instructions that, when executed by the at least one processor, cause the apparatus at least to determine a configuration for artificial intelligence / machine learning (AI / ML) beam management to be performed by a user device and determine one or more delay requirements associated with the AI / ML beam management. The apparatus may also be caused to determine a time instance based on the determined one or more delay requirements associated with the AI / ML beam management and transmit, to the user device, the configuration for AI / ML beam management. The configuration may indicate a reporting frequency for the AI / ML beam management as semi-persistent reporting. The apparatus may further be caused to receive, via semi-persistent reporting from the user device during the time instance, a report indicating one or more inference results of the AI / ML beam management.

[0007] Various exemplary embodiments may provide an apparatus comprising at least one processor and at least one memory storing instructions that, when executed by the at least one processor, cause the apparatus at least to receive, from a network entity, a configuration for artificial intelligence / machine learning (AI / ML) beam management. The configuration may indicate a reporting frequency for AI / ML beam management as aperiodic reporting. The apparatus may also be caused to determine one or more delay requirements associated with the AI / ML beam management and determine a time instance based on the determined one or more delay requirements associated with the AI / ML beam management. The apparatus may be further caused to perform the AI / ML beam management and transmit, via aperiodic reporting to the network entity during the time instance, a report indicating one or more inference results of the AI / ML beammanagement.

[0008] Some exemplary embodiments may provide an apparatus comprising at least one processor and at least one memory storing instructions that, when executed by the at least one processor, cause the apparatus at least to determine a configuration for artificial intelligence / machine learning (AI / ML) beam management to be performed by a user device and determine one or more delay requirements associated with the AI / ML beam management. The apparatus may also be caused to determine a time instance based on the determined one or more delay requirements associated with the AI / ML beam management and transmit, to the user device, the configuration for AI / ML beam management. The configuration may indicate a reporting frequency for the AI / ML beam management as aperiodic reporting. The apparatus may further be caused to receive, via aperiodic reporting from the user device during the time instance, a report indicating one or more inference results of the AI / ML beam management.

[0009] Various exemplary embodiments may provide one or more methods which include receiving, by an apparatus from a network entity, a configuration for artificial intelligence / machine learning (AI / ML) beam management. The configuration may indicate a reporting frequency for the AI / ML beam management as periodic reporting. The method may also include determining one or more delay requirements associated with the AI / ML beam management, determining a time instance based on the determined one or more delay requirements associated with the AI / ML beam management, performing the AI / ML beam management, and transmitting, via periodic reporting to the network entity during the time instance, a report indicating one or more inference results of the AI / ML beam management.

[0010] Certain exemplary embodiments may provide a method including determining a configuration for artificial intelligence / machine learning (AI / ML) beam management to be performed by a user device and determining one or more delay requirements associated with the AI / ML beam management. The method may also include determining a time instance based on the determined one or more delay requirements associated with the AI / ML beam management and transmitting to the user device, the configuration for AI / ML beam management. The configuration may indicate a reporting frequency for the AI / ML beam management as periodic reporting. The method may further include receiving, via periodic reporting from the user device during the time instance, a report indicating one or more inference results of the AI / ML beam management.

[0011] Some exemplary embodiments may provide a method including receiving, from a network entity, a configuration for artificial intelligence / machine learning (AI / ML) beam management. The configuration may indicate a reporting frequency for the AI / ML beam management as semi-persistent reporting. The method may also include determining one or more delay requirements associated with the AI / ML beam management, determining a time instance based on the determined one or more delay requirements associated with the AI / ML beam management, performing the AI / ML beam management, and transmitting, via semi-persistent reporting to the network entity during the time instance, a report indicating one or more inference results of the AI / ML beam management.

[0012] Certain exemplary embodiments may provide a method including determining a configuration for artificial intelligence / machine learning (AI / ML) beam management to be performed by a user device and determining one or more delay requirements associated with the AI / ML beam management. The method may also include determining a time instance based on the determined one or more delay requirements associated with the AI / ML beam management and transmitting, to the user device, the configuration for AI / ML beam management. The configuration may indicate a reporting frequency for the AI / ML beam management as semi-persistent reporting. The method may further include receiving, via semi-persistent reporting from the user device during the time instance, a report indicating one or more inference results of the AI / ML beam management.

[0013] Various exemplary embodiments may provide a method including receiving, from a network entity, a configuration for artificial intelligence / machine learning (AI / ML) beam management. The configuration may indicate a reporting frequency for AI / ML beam management as aperiodic reporting. The method may also include determining one or more delay requirements associated with the AI / ML beam management and determining a time instance based on the determined one or more delay requirements associated with the AI / ML beam management. The method may further include performing the AI / ML beam management and transmitting, via aperiodic reporting to the network entity during the time instance, a report indicating one or more inference results of the AI / ML beam management.

[0014] Some exemplary embodiments may provide a method including determining a configuration for artificial intelligence / machine learning (AI / ML) beam management to be performed by a user device and determining one or more delay requirements associated with the AI / ML beam management. The method may also include determining a time instance based on the determined one or more delay requirements associated with the AI / ML beam management and transmitting, to the user device, the configuration for AI / ML beam management. The configuration may indicate a reporting frequency for the AI / ML beam management as aperiodic reporting. The method may further include receiving, via aperiodic reporting from the user device during the time instance, a report indicating one or more inference results of the AI / ML beam management.

[0015] Certain exemplary embodiments may provide a non-transitory computer-readable storage medium comprising computer-executable instructions that, when executed by an apparatus, cause the apparatus to perform any one or more of the methods described herein.

[0016] Some exemplary embodiments may provide one or more computer programs including instructions stored thereon for performing one or more of the methods described herein.

[0017] Certain exemplary embodiments may provide one or more apparatuses including one or more circuitry configured to perform one or more of the methods described herein.

[0018] Various exemplary embodiments may provide one or more apparatuses including one or more means configured for performing one or more of the methods described herein.BRIEF DESCRIPTION OF THE DRAWINGS:

[0019] For proper understanding of exemplary embodiments, reference should be made to the accompanying drawings, as follows:

[0020] FIG.1 illustrates an example of a signal diagram, according to various exemplary embodiments;

[0021] FIG. 2 illustrates an example of a flow diagram of a method, according to some exemplary embodiments;

[0022] FIG.3 illustrates an example of a flow diagram of another method, according to various exemplary embodiments;

[0023] FIG.4 illustrates an example of a flow diagram of a further method, according to certain exemplary embodiments;

[0024] FIG. 5 illustrates an example of a flow diagram of an additional method, according to various exemplary embodiments;

[0025] FIG. 6 illustrates an example of a flow diagram of a method, according to some exemplary embodiments;

[0026] FIG. 7 illustrates an example of a flow diagram of a method, according to certain exemplary embodiments; and

[0027] FIG.8 illustrates a set of apparatuses, according to various exemplary embodiments. DETAILED DESCRIPTION:

[0028] It will be readily understood that the components of certain exemplary embodiments, as generally described and illustrated in the figures herein, may be arranged and designed in a wide variety of different configurations. The following is a detailed description of some exemplary embodiments of systems, methods, apparatuses, and non-transitory computer program products for implementing delay aware life cycle management operations for beam prediction. Although the devices discussed below and shown in the figures refer to 6G / 5G or Next Generation NodeB (gNB) devices and UE devices, this disclosure is not limited to only gNBs and UEs.

[0029] It may be readily understood that the components of certain exemplary embodiments, as generally described and illustrated in the figures herein, may be arranged and designed in a wide variety of different configurations. Different reference designations from multiple figures may be used out of sequence in the description, to refer to a same element to illustrate their features or functions. If desired, the different functions or procedures discussed herein may be performed in a different order and / or concurrently with each other. Furthermore, if desired, one or more of the described functions or procedures may be optional or may be combined. As such, the following description should be considered as illustrative of the principles and teachings of certain exemplary embodiments, and not in limitation thereof.

[0030] 3rdGeneration Partnership Project (3GPP) may provide specifications for implementing artificialintelligence / machine learning (AI / ML). Data collection may enable ML models in an NR air interface to be trained by generating a training dataset from a variety of data sources within and outside of a 6G / 5G network. A significant amount of data related to network and service events and status may be able to be analyzed and processed for various implementations.

[0031] AI / ML based beam management may be implemented for spatial domain beam prediction and / or time domain beam prediction. Spatial beam prediction may be used to predict one or more optimal transmit (Tx) and / or receive (Rx) beams in different spatial locations. Time-domain beam predictions may be used to predict a most likely beam to use for the next one or more time instants. This time-domain beam prediction may even predict the beam to be used in the spatial domain.

[0032] A life cycle management (LCM) scheme may be implemented to manage and optimize the AI / ML based beam management. LCM may provide procedures, for example, for selection, activation, deactivation, switching, and / or fallback of one or more AI / ML models and applications. LCM may be applied to facilitate model training, inference, performance monitoring, and / or data collection for UE-side models and / or network side models.

[0033] For AI / ML beam management, the LCM operations, such as, for example, activation, deactivation, switching, and / or fall-back (e.g., changing from AI / ML beam prediction to legacy beam prediction) may be supported by channel state information (CSI) reporting configurations. For example, one or more of the LCM operations may be triggered, for example, based on the performance monitoring output of an AI / ML model. The network or network entity may be able to configure CSI reporting configurations to enable beam predictions via AI / ML beam management for spatial beam prediction and / or time-domain beam prediction. The CSI reports for spatial beam prediction and / or time-domain beam prediction may be based on, for example, periodic, semi-persistent, and / or aperiodic CSI reporting.

[0034] Various exemplary embodiments may recognize that it is desirable for LCM operations in AI / ML beam management to account for latency / delay. Delay may be caused by a variety of factors and / or circumstances, such as, for example, whether CSI configuration (ResourceSet) to be used is already available at the UE, the type of CSI reporting (periodic, semi-persistent, aperiodic), the availability of previous or current measurements to be used as an input to the model for functionality inference or prediction, availability of the necessary model(s) at the UE, and / or time to substitute one model / functionality with another (e.g., the UE may have specialized AI-hardware accelerators to run the inference, which may require loading of the model). It may be desirable for the UE and the network or network entity to be informed of the delay(s) to allow the network to determine when to schedule grants for the reporting of measurements, predictions / inferences, and / or data.

[0035] Various exemplary embodiments may provide technological advantages to address the above- mentioned delays and implement one or more procedures for determining delay in LCM operations and accounting for the delay to optimize performance. Certain exemplary embodiments may provide forspecifying the amount of time it takes for the UE to report predictions or inference results following one or more changes indicated by LCM operation(s). The UE and the network may be informed of the delay(s) to allow the network to determine when to schedule grants for the reporting of measurements, predictions / inferences, and / or data.

[0036] Certain exemplary embodiments may provide for determining and managing delay due to, for example, the periodicity of CSI reporting. The periodicity of the CSI reporting may be, for example, periodic, semi-persistent, and / or aperiodic. For periodic CSI reporting of beam prediction, radio resource control (RRC) may be used for triggering reporting. Reconfiguration may be used to change from a different CSI report to periodic CSI reporting. For semi-persistent CSI reporting of beam prediction, LCM operations, such as selection, activation / deactivation, switching, and fallback, of different CSI reports may be operated with or controlled by medium access control control element (MAC-CE) signaling. For aperiodic CSI reporting of beam prediction, downlink control information (DCI) based triggering for CSI reporting may be used and the DCI signaling may be used for selection, activation, and / or deactivation of CSI reports. The fallback LCM operation for CSI reporting from AI / ML beam management to legacy beam management may be triggered upon request or signaling.

[0037] Various exemplary embodiments may provide that delays due to periodic CSI reporting may be the same or different depending on the type of LCM operation (e.g., activation, deactivation, switching, or fall- back). A delay from activation and / or deactivation LCM operations may be referred to as ^^^^^^^^^^^. Examples of delay due to activation / deactivation LCM operations may include delay from the time to performRRC reconfiguration (^^^^^^^^^^^^^^^^^^^ + ^^^^^^^^^^^^^^^^^^^_^^^^^^^^), delay from the time for theUE to perform processing related to generating a capability report (^^^^^^^^^^^^^_^^^^^^^^^^_^^^^^^), delay related to obtaining or determining an associated identifier (ID) (^^^^^^^^^^^_^^), which is an ID associated with the CSI reporting, and / or delay due to the time for CSI reference signal (CSI-RS) measurements to be performed, which may also be referred to as a CSI-RS measurement period (^^^ ^!^"_#^^^_"^^^^^_^!^ ^!).

[0038] A delay from a switching LCM operation during periodic CSI reporting may be referred to as ^^$^^^%^^^. Examples of delay due to switching LCM operations may include delay from the time to performa new RRC reconfiguration (^^^^^^^^^^^^^^^^^^^ + ^^^^^^^^^^^^^^^^^^^_^^^^^^^^), delay from the timefor the UE to perform processing related to generating a capability report (^^^^^^^^^^^^^_^^^^^^^^^^_^^^^^^), delay related to obtaining or determining an associated IDthe CSI reporting, and / or delay due to the time for CSI-RS measurements to be performed, which may also be referred to as a CSI-RS measurement period, (^^^ ^!^"_#^^^_"^^^^^_^!^ ^!).

[0039] A delay from a fall-back LCM operation during periodic CSI reporting, which may be changing from AI / ML to legacy (non-AI / ML) beam management, may be referred to as ^&^^^^^^'. Examples of delay due to fall-back LCM operations may include delay from the time to performa new RRC reconfiguration (^^^^^^^^^^^^^^^^^^^ + ^^^^^^^^^^^^^^^^^^^_^^^^^^^^), delay from the timefor the UE to perform processing related to generating a capability report (^^^^^^^^^^^^^_^^^^^^^^^^_^^^^^^), and / or delay due to the time for CSI-RS measurements to be performed, which may also be referred to as a CSI-RS measurement period (^^^ ^!^"_#^^^_^!^ ^!_^^^^^^).

[0040] Some exemplary embodiments may provide that delays due to semi-persistent CSI reporting may be the same or different depending on the type of LCM operation. A delay from activation and / or deactivation LCM operations may be referred to as ^^^^^^^^^^^. Examples of delay due to activation / deactivation LCM operations may include delay from the time to activate one or more semi-persistent CSI resource sets (^^^^^^^^^_^^^^_^!^^^^^^^^^^^^), delay from the time to select one or more semi-persistent CSI resource sets (^^^^^^^^^^_^^^^_^!^^^^^^^^^^^^), delay from the time for the UE to perform processing related to generating a capability report (^^^^^^^^^^^^^_^^^^^^^^^^_^^^^^^), delay related to obtaining or determining an associated ID (^^^^^^^^^^^_^^), and / or delay due to the time for a CSI-RS measurement period (^^^ ^!^"_#^^^_"^^^^^_^!^ ^!).

[0041] A delay from a switching LCM operation during semi-persistent CSI reporting may be referred to as ^^$^^^%^^^. Examples of delay due to switching LCM operations may include delay from the time to select new one or more semi-persistent CSI resource sets (^^^^^^^^^^_^^^^_^!^^^^^^^^^^^^), delay from the time for the UE to perform processing related to generating a capability report (^^^^^^^^^^^^^_^^^^^^^^^^_^^^^^^), delay related to obtaining or determining an associated ID (^^^^^^^^^^^_^^), and / or delay due to the time for a CSI-RS measurement period (^^^ ^!^"_#^^^_"^^^^^_^!^ ^!).

[0042] A delay from a fall-back LCM operation during semi-persistent CSI reporting may be referred to as ^&^^^^^^'. Examples of delay due to fall-back LCM operations may include delay from the time to activate one or more semi-persistent CSI resource sets (^^^^^^^^^_^^^^_^!^^^^^^^^^^^^), delay from the time to select one or more semi-persistent CSI resource sets (^^^^^^^^^^_^^^^_^!^^^^^^^^^^^^), delay from the time for the UE to perform processing related to generating a, and / or delay due to the time for a CSI-RS measurement period (^^^ ^!^"_#^^^_^!^ ^!_^^^^^^).

[0043] Certain exemplary embodiments may provide that delays due to aperiodic CSI reporting may be the same or different depending on the type of LCM operation. The delay from activation and / or deactivation LCM operations may be referred to as ^^^^^^^^^^^. Examples of delay due to activation / deactivation LCM operations may include delay from the time for the UE to perform processing related to generating a capabilityreport ( ^^^^^^^^^^^^^_^^^^^^^^^^_^^^^^^ ), delay related to obtaining or determining an associated ID( ^^^^^^^^^^^_^^ ), and / or delay due to the time for a CSI-RS measurement period(^^^ ^!^"_#^^^_"^^^^^_^!^ ^!).

[0044] A delay from a switching LCM operation during aperiodic CSI reporting may be referred to as^^$^^^%^^^. Examples of delay due to switching LCM operations may include delay from the time for the UE to perform processing related to generating a capability report (^^^^^^^^^^^^^_^^^^^^^^^^_^^^^^^), delay related to obtaining or determining an associated ID (^^^^^^^^^^^_^^), and / or delay due to the time for a CSI- RS measurement period, (^^^ ^!^"_#^^^_"^^^^^_^!^ ^!).

[0045] A delay from a fall-back LCM operation during aperiodic CSI reporting may be referred to as ^&^^^^^^'. Examples of delay due to fall-back LCM operations may include delay for the time for the UE to perform processing related to generating a capability report (^^^^^^^^^^^^^_^^^^^^^^^^_^^^^^^), and / or delay due to the time for CSI-RS measurements to be performed, which may also be referred to as a CSI-RS measurement period (^^^ ^!^"_#^^^_^!^ ^!_^^^^^^).

[0046] Various exemplary embodiments may provide that the activation LCM operation may be the first operation to be performed when a UE receives from the network an inference configuration for performing an inference with an AI / ML model available at the UE. After the activation operation, other operations, e.g., deactivation, switching and / or fallback, may be performed.

[0047] For beam prediction reporting enabled by semi-persistent CSI reporting, which may use a radio network temporary indicator (SP-CSI-RNTI), the network or network entity, such as a gNB, may trigger by DCI to activate semi-persistent CSI reporting and may also trigger MAC-CE to select a semi-persistent resource set listed from RRC. The semi-persistent CSI reporting and MAC-CE may cause delay in LCM operations. For beam prediction reporting enabled by aperiodic CSI reporting, a DCI may be used to trigger a state associated with one or more CSI reporting configurations. The network may transmit DCI to enable two CSI reporting configuration, such as, for example, one for beam prediction resource set and another one for beam measurements resource set. For beam prediction reporting enabled by periodic CSI reporting, semi-persistent CSI reporting, or aperiodic CSI reporting, the beam reporting may be performed after any other delay elements, such as, for example, up to 8 beams reporting.

[0048] Various exemplary embodiments may provide that AI / ML beam prediction reporting enabled by periodic CSI-report may provide multiple LCM-related latency / delay conditions or requirements. For the LCM operations of activating or switching an inference operation of AI / ML-enabled beam prediction via periodic CSI reporting, a total delay may be defined as a delay requirement. The total delay may be based on one or a combination of multiple delays. For example, a delay due to RRC latency, e.g., RRC-reconfiguration andRRC reconfiguration complete ( ^^^^^^^^^^^^^^^^^^^ + ^^^^^^^^^^^^^^^^^^^_^^^^^^^^ ) may beconsidered. The network (e.g., gNB) may send RRC (re)configuration to a UE and the UE may send RRC a (re)configuration complete message back to the network. As an example from 3GPP specifications, this delay component may be obtained from TS 38.331.

[0049] Another example delay may be a delay from a processing time in which the UE takes to perform processes related to a UE capability report, such as generating the capability report. The processing timemay be related to the quantities of the inference report, sub-use cases or feature groups, capability model- related pre-processing time, and / or memory limitations on storing models associated with beam prediction. This delay from processing time by the UE may be referred to as ^^^^^^^^^^^^^_^^^^^^^^^^_^^^^^^. Certain exemplary embodiments may provide various examples of delays caused by the UE processing time. For example, when the UE performs Top-K beam IDs prediction, the UE may have ^^ms preparation time, or when the UE performs Top-K beam IDs and Top-K layer 1 reference signal received power (L1-RSRP) prediction, the UE may have ^)ms preparation time. Top-K may, for example,to one or more strongest K beams, where K=1 by default and / or K > 1. Another example of delay from UE processing time may be delay from inference operation(s), such as CSI computation time ^^!^ ^^^^^^^^^^^ ^^^^^. When the UE performs beam prediction for the spatial domain, the UE processing time may be *^ms and when the UE performs beam prediction for the time domain, the UE processing time may be ms.

[0050] Another example of delay from UE processing time may be on whether the AI / ML model / functionality is already in use or was in use recently, such that the AI / ML model is in the buffer memory, which may result in different UE processing time delays in following different cases. When the AI / ML model / functionality is activated for the first time, a higher delay may occur to load the AI / ML model and for data pre-processing and post processing, as compared to a situation in which the AI / ML model / functionality has been activated recently to be in the buffer memory. When the AI / ML model / functionality is to be switched from one model / functionality to another model / functionality, which may result in unloading of current model / functionality, a delay time may be caused due to loading the new model and changing data pre- processing and post processing. This delay time may further depend on the size of the model and data processing used for switching. When the AI / ML model / functionality is deactivated (e.g., fallback to legacy), then a different value of UE processing delay, such as comparatively less delay, may occur to unload the model / functionality and switch from AI / ML data pre-processing and post processing to legacy data pre- processing and post processing. When the AI / ML model / functionality is activated after being deactivated recently, such that the model / functionality is still in the buffer memory of the UE, the UE processing delay may be shorter than when the model / functionality is not in the buffer memory.

[0051] Some exemplary embodiments may provide that delay from UE processing time may be associated with the AI / ML model / functionality to be activated / deactivated / switched. Certain AI / ML models / functionalities may be greater in size and may need more data pre-processing and post processing as compared to other models / functionalities. For example, spatial beam prediction may have a different model size and may use different data processing than time domain beam prediction.

[0052] Another example delay may be a delay from a processing time related to determining or obtaining the associated ID(s) related to resource sets. For example, a non-zero power CSI-RS resource set, such as NZP-CSI-RS-ResourceSet of Set A and NZP-CSI-RS-ResourceSet / CSI-SSB-ResourceSet of Set B may benew (i.e., not used in the past), then the UE may be defined to consider one value for ^^^^^^^^^^^_^^, as compared to when the associated-ID(s) related to NZP-CSI-RS-ResourceSet of Set A and NZP-CSI-RS- ResourceSet / CSI-SSB-ResourceSet of Set B have been used in the past, in which the UE may be defined to consider a different value for ^^^^^^^^^^^_^^. The terms “used in the past” or “not used in the past” may be defined based on a time window or timer considered by the UE, where the time window may be defined with reference to a time instance when there an associated ID is configured / indicated to the UE. For example,^^^^^^^^^^^+, = 0 may be set when the associated ID is used in the past; otherwise, ^^^^^^^^^^^+, > 0.As another example, ^^^^^^^^^^^+,may include delays associated with model transfer or model download to the UE.

[0053] A further example delay may be a delay from a processing time related to a CSI-RS measurement period (^^^ ^!^"_#^^^_"^^^^^_^!^ ^!). For example, in spatial beam prediction, the delay may be based on a measured L1-RSRP of the current time instance, and in time domain beam prediction, the delay may be based on a measured observation window, which may be a measured historical L1-RSRP.

[0054] Certain exemplary embodiments may provide a total delay that may be defined for deactivating or changing to fallback to non-AI / ML operation for beam measurement via periodic CSI reporting. The delay related to RRC latency, e.g., RRC-reconfiguration and RRC reconfiguration complete, may be considered in the total delay. The network may deactivate the current periodic CSI-report (current model / functionality) and may configure the UE to fallback to legacy beam management or prediction. The network may send a new RRC-reconfiguration to the UE to operate in legacy mode. Delay may also occur from UE processing time related to a capability report (^^^^^^^^^^^^^_^^^^^^^^^^), such as, for example, the UE processing time to report up to Top-4 beams in legacy mode. Delay may further occur from the time of the CSI-RS measurement period (^^^ ^!^"_#^^^_^!^ ^!), in which the delay of the measurement period in legacy may be considered.

[0055] Some exemplary embodiments may provide that when periodic CSI reporting is enabled to support AI / ML beam prediction, a delay for activation LCM operations may be defined as: ^^^^^^^^^^^ = ^^^^1234567891:;745 + ^^^^^^^^^^^^^^^^^^^_^^^^^^^^ +^!

[0056] When periodic CSI reporting is enabled to support AI / ML beam prediction, a delay for a switching LCM operation to switch from a current CSI-ReportConfig to another current CSI-ReportConfig may be defined as: ^^$^^^%^^^ = ^^^^1234567891:;745 + ^^^^^^^^^^^^^^^^^^^_^^^^^^^^+ ^^^^^^^^^^^_^^^!

[0057] When periodic CSI reporting is enabled to support AI / ML beam prediction, a delay for a fallback LCM operation to fallback from a current CSI-ReportConfig to a legacy beam prediction mode may be defined as:^&^^^^^^' = ^^^^^^^^^^^^^^^^^^^ + ^^^^^^^^^^^^^^^^^^^_^^^^^^^^ + ^^^^^^^^^^^^^_^^^^^^^^^^_^^^^^^+ ^^^^^^^^^^^^^_^^^^^^^^^^_^^^^^^

[0058] Various exemplary embodiments may provide AI / ML beam prediction reporting enabled by semi- persistent CSI reporting. For activating or switching inference LCM operations of AI / ML-enabled beam prediction via semi-persistent CSI reporting, the delays may be the same as the delays for periodic CSI reporting with some exceptions. For example, delay from activating a semi-persistent CSI resource set triggered by DCI (^^^^^^^^^_^^^^_^!^^^^^^^^^^^^) and delay for selecting of semi-persistent CSI resource set triggered by MAC-CE (^^^^^^^^^^_^^^^_^!^^^^^^^^^^^^). These two delays may be related to network triggering DCI to and trigger MAC-CE to select a semi-persistentresource set.

[0059] Some exemplary embodiments may provide that when the UE is configured to switch from one model / functionality of AI / ML beam prediction supported by semi-persistent CSI reporting to another model / functionality of AI / ML beam prediction supported by semi-persistent CSI reporting, delay may be from the network, e.g., gNB, selecting of one or more semi-persistent CSI resource sets (^^^^^^^^^^_^^^^_^!^^^^^^^^^^^^), where the network may configure the UE to deactivate a current semi- persistent CSI-ReportConfig associated with a current model / functionality and trigger the UE to switch to a new semi-persistent CSI-ReportConfig for a new model / functionality via MAC-CE.

[0060] Certain exemplary embodiments may provide that for deactivating or changing to fallback to non- AI / ML operations for beam measurement via semi-persistent CSI reporting, a total delay may be defined as the delay requirement. For example, delay may result from activating a semi-persistent CSI resource set triggered by DCI (^^^^^^^^^_^^^^_^!^^^^^^^^^^^^). The network may deactivate the current semi-persistent CSI-ReportConfig for the current model / functionality and may trigger the UE to activate a semi-persistent CSI-ReportConfig for legacy beam management. As another example, delay may be due to selecting a semi- persistent CSI resource set in legacy beam prediction mode triggered by MAC-CE (^^^^^^^^^^_^^^^_^!^^^^^^^^^^^^).

[0061] Various exemplary embodiments may provide that when semi-persistent CSI reporting is enabled to support AI / ML beam prediction, a delay for activation LCM operations, such as to activate AI / ML beam prediction for spatial beam prediction and / or time domain beam prediction, may be defined as: ^^^^^^^^^^^ = ^^^^^^^^^_^^^^_^!^^^^^^^^^^^^ + ^^^^^^^^^^_^^^^_^!^^^^^^^^^^^^+ ^^^^^^^^^^^^^_^^^^^^^^^^_^^^^^^ + ^^^ ^!^"_#^^^_"^^^^^_^!^ ^! + ^^^^^^^^^^^_^^

[0062] When semi-persistent CSI reporting is enabled to support AI / ML beam prediction, a delay for switching LCM operations, such as switching from a current CSI-ReportConfig to another CSI-ReportConfig, may be defined as:^^$^^^%^^^ = ^^^^^^^^^^_^^^^_^!^^^^^^^^^^^^+^^^^^^^^^^^^^_^^^^^^^^^^_^^^^^^ + ^^^^^^^^^^^_^^+ ^^^ ^!^"_#^^^_"^^^^^_^!^ ^!

[0063] When semi-persistent CSI reporting is enabled to support AI / ML beam prediction, a delay for fallback LCM operations, such as fallback from the current CSI-ReportConfig to a legacy reporting, may be defined as: ^&^^^^^^' = ^^^^^^^^^_^^^^_^!^^^^^^^^^^^^ + ^^^^^^^^^^_^^^^_^!^^^^^^^^^^^^+ ^^^^^^^^^^^^^_^^^^^^^^^^_^^^^^^ + ^^^^^^^^^^^^^_^^^^^^^^^^_^^^^^^

[0064] Certain exemplary embodiments may provide AI / ML beam prediction reporting enabled by aperiodic CSI reporting. For activating or switching inference LCM operations of AI / ML-enabled beam prediction via aperiodic CSI reporting, the delays may be the same as the delays for periodic CSI reporting with some exceptions. For example, the delay may be due to activating an aperiodic CSI resource set triggered by DCI (^^^^^^^^^_^"_^!^^^^^^^^^^^^), where a single DCI may be used for the trigger.

[0065] When a UE is configured to switch from one model / functionality of AI / ML beam prediction supported by aperiodic CSI reporting to another model / functionality of AI / ML beam prediction supported by aperiodic CSI reporting, delay may be from the network, e.g., gNB, selecting of one or more aperiodic CSI resource sets by sending a new CSI-ResourceConfig using DCI, in which the delay may be ^^^^^^^^^_^"_^!^^^^^^^^^^^^. For deactivating or changing to a fallback operation to fallback from AI / ML beam measurement / prediction to non-AI / ML operations for beam measurement via aperiodic CSI reporting, a total delay may be based on the time to activate one or more semi-persistent CSI resource sets triggered by DCI (^^^^^^^^^_^"_^!^^^^^^^^^^^^).

[0066] Various exemplary embodiments may provide that when aperiodic CSI reporting is enabled to support AI / ML beam prediction, a delay for activation LCM operations, such as the UE activating AI / ML beam prediction, may be defined as: ^^^^^^^^^^^ = ^^^^^^^^^_^"_^!^^^^^^^^^^^^ + ^^^^^^^^^^^^^_^^^^^^^^^^_^^^^^^ + ^^^^^^^^^^^_^^+ ^^^ ^!^"_#^^^_"^^^^^_^!^ ^!

[0067] When aperiodic CSI reporting is enabled to support AI / ML beam prediction, a delay for switching LCM operations, such as the UE switching from current CSI-ReportConfig to another CSI-ReportConfig, may be defined as: ^^$^^^%^^^ = ^^^^^^^^^_^"_^!^^^^^^^^^^^^ + ^^^^^^^^^^^^^_^^^^^^^^^^_^^^^^^ + ^^^^^^^^^^^_^^+ ^^^ ^!^"_#^^^_"^^^^^_^!^ ^!

[0068] When aperiodic CSI reporting is enabled to support AI / ML beam prediction, a delay for fallback LCM operations, such as the UE performing fallback from a current CSI-ReportConfig to a legacy reporting, may be defined as:^^^^^^^^' = ^^^^^^^^^_^"_^!^^^^^^^^^^^^ + ^^^^^^^^^^^^^_^^^^^^^^^^_^^^^^^+ ^^^ ^!^"_#^^^_"^^^^^_^!^ ^!

[0069] Various exemplary embodiments may provide technological advantages to implement one or more procedures for determining delay in LCM operations and accounting for the delay to optimize performance. Certain exemplary embodiments may provide for specifying the amount of time it takes for the UE to report predictions or inference results following one or more changes indicated by LCM operation(s). The UE and the network may be informed of the delay(s) to allow the network to determine when to schedule grants for the reporting of measurements, predictions / inferences, and / or data.

[0070] FIG.1 illustrates an example of a signal diagram, according to certain exemplary embodiments. The signal diagram shows signaling between a UE 10 and a network 20, such as, for example, a gNB 20. At 101, the network 20 may determine a configuration for AI / ML beam management to be performed by the UE 10. At 102, the network 20 may determine one or more delay requirements associated with the AI / ML beam management, and at 103, the network 20 may determine a time instance based on the determined one or more delay requirements.

[0071] At 104, the network 20 may transmit, to the UE 10, a configuration for AI / ML beam management. At 105, the UE 10 may determine one or more delay requirements associated with the AI / ML beam management, and at 106, the UE 10 may determine a time instance based on the determined one or more delay requirements. At 107, the UE 10 may perform the AI / ML beam management operations / functions, and at 108, the UE 10 may provide, to the network 20, one or more inference results outputted from the AI / ML beam management operations / functions. The one or more inference results may be provided during the determined time instance.

[0072] FIG. 2 illustrates an example flow diagram of a method, according to certain exemplary embodiments. In an example embodiment, the method of FIG.2 may be performed by a user device in a 3GPP system, such as LTE, 5G-NR, or 6G. For instance, in an exemplary embodiment, the method of FIG. 2 may be performed by a user apparatus device or user equipment, such as a UE, similar to apparatus 810 illustrated in FIG.8.

[0073] According to various exemplary embodiments, the method of FIG.2 may include, at 210, receiving, from a network entity, a configuration for artificial intelligence / machine learning (AI / ML) beam management. The configuration may indicate a reporting frequency for the AI / ML beam management as periodic reporting. At 220, the method may include determining one or more delay requirements associated with the AI / ML beam management, and at 230, determining a time instance based on the determined one or more delay requirements associated with the AI / ML beam management. The method may further include, at 240, performing the AI / ML beam management, and at 250, transmitting, via periodic reporting to the network entity during the time instance, a report indicating one or more inference results of the AI / ML beam management.

[0074] Certain exemplary embodiments may provide that the configuration for AI / ML beam managementalso indicates a life cycle management operation. The life cycle management operation may include at least one of the following: an activation operation for AI / ML inference, a deactivation operation for AI / ML inference, a switching operation from a first AI / ML inference to a second AI / ML inference, or a fallback operation to change from an AI / ML inference to a legacy non-AI / ML inference. The one or more delay requirements for the activation operation or the deactivation operation may include delay from performing radio resource control reconfiguration, delay from processing by the apparatus of a capability report of the apparatus, delay from obtaining an associated reporting identifier, and / or delay from a measurement period of a channel state information reference signal. The one or more delay requirements for the switching operation may include delay from performing a new radio resource control reconfiguration, which is different from a prior radio resource control configuration, delay from processing by the apparatus of a capability report of the apparatus, delay from obtaining an associated reporting identifier, and / or delay from a measurement period of a channel state information reference signal.

[0075] Some exemplary embodiments may provide that the measurement period of the channel state information reference signal is a time period for measuring a power level of the channel state information reference signal. The one or more delay requirements may include a processing time for the apparatus to generate a report of the inference results. A delay in the processing time may include at least one of a preparation time for the apparatus to perform beam identification prediction or a preparation time for the apparatus to perform beam identification prediction and reference signal power prediction. The delay in the processing time may include a processing time for obtaining the one or more inference results of the AI / ML beam management for beam prediction. The delay in the processing time may include a buffering time for an AI / ML functionality of the AI / ML beam management. The buffering time may be an amount of time to load an AI / ML model and for pre and post processing of data processed by the AI / ML model.

[0076] FIG. 3 illustrates an example flow diagram of a method, according to certain exemplary embodiments. In an example embodiment, the method of FIG. 3 may be performed by a network element / entity, or a group of multiple network entities in a 3GPP system, such as LTE, 5G-NR, or 6G. For instance, in an exemplary embodiment, the method of FIG.3 may be performed by a network node or network entity, such as a gNB, similar to apparatus 820 illustrated in FIG.8.

[0077] According to various exemplary embodiments, the method of FIG.3 may include, at 310, determining a configuration for artificial intelligence / machine learning (AI / ML) beam management to be performed by a user device. At 320, the method may include determining one or more delay requirements associated with the AI / ML beam management, and at 330, determining a time instance based on the determined one or more delay requirements associated with the AI / ML beam management. The method may further include, at 340, transmitting to the user device, the configuration for AI / ML beam management. The configuration may indicate a reporting frequency for the AI / ML beam management as periodic reporting. At 350, the method may include receiving, via periodic reporting from the user device during the time instance, a report indicatingone or more inference results of the AI / ML beam management.

[0078] Certain exemplary embodiments may provide the configuration for AI / ML beam management also indicates a life cycle management operation. The life cycle management operation may include at least one of the following: an activation operation for AI / ML inference, a deactivation operation for AI / ML inference, a switching operation from a first AI / ML inference to a second AI / ML inference, or a fallback operation to change from an AI / ML inference to a legacy non-AI / ML inference.

[0079] Some exemplary embodiments may provide that the one or more delay requirements for the activation operation or the deactivation operation comprises at least one of delay from the user device performing radio resource control reconfiguration, delay from processing by the user device of a capability report of the user device, delay from the user device obtaining an associated reporting identifier, and / or delay from a measurement period of a channel state information reference signal. The one or more delay requirements for the switching operation may include at least one of delay from the user device performing a new radio resource control reconfiguration, which is different from a prior radio resource control configuration, delay from processing by the user device of a capability report of the user device, delay from the user device obtaining an associated reporting identifier, and / or delay from a measurement period of a channel state information reference signal.

[0080] Certain exemplary embodiments may provide that the measurement period of the channel state information reference signal is a time period for measuring a power level of the channel state information reference signal. The one or more delay requirements may include a processing time for the user device to generate a report of the inference results. A delay in the processing time may include at least one of a preparation time for the user device to perform beam identification prediction or a preparation time for the user device to perform beam identification prediction and reference signal power prediction. The delay in the processing time may include a processing time for the user device to obtain the one or more inference results of the AI / ML beam management for beam prediction. The delay in the processing time may include a buffering time for an AI / ML functionality of the AI / ML beam management. The buffering time may be an amount of time for the user device to load an AI / ML model and for pre and post processing of data processed by the AI / ML model.

[0081] FIG. 4 illustrates an example flow diagram of a method, according to certain exemplary embodiments. In an example embodiment, the method of FIG.4 may be performed by a user device in a 3GPP system, such as LTE, 5G-NR, or 6G. For instance, in an exemplary embodiment, the method of FIG. 4 may be performed by a user apparatus / device or user equipment, such as a UE, similar to apparatus 810 illustrated in FIG.8.

[0082] According to various exemplary embodiments, the method of FIG.4 may include, at 410, receiving, from a network entity, a configuration for artificial intelligence / machine learning (AI / ML) beam management. The configuration may indicate a reporting frequency for the AI / ML beam management as semi-persistentreporting. The method may also include, at 420, determining one or more delay requirements associated with the AI / ML beam management, and at 430, determining a time instance based on the determined one or more delay requirements associated with the AI / ML beam management. The method may further include, at 440, performing the AI / ML beam management, and at 450, transmitting, via semi-persistent reporting to the network entity during the time instance, a report indicating one or more inference results of the AI / ML beam management.

[0083] Certain exemplary embodiments may provide the configuration for AI / ML beam management also indicates a life cycle management operation comprising at least one of the following: an activation operation for AI / ML inference, a deactivation operation for AI / ML inference, a switching operation from a first AI / ML inference to a second AI / ML inference, or a fallback operation to change from an AI / ML inference to a legacy non-AI / ML inference. The one or more delay requirements for the activation operation or the deactivation operation may include at least one of: delay for activating a semi-persistent channel state information resource set, delay for selecting the semi-persistent channel state information resource set, delay from processing by the apparatus of a capability report of the apparatus, delay from obtaining an associated reporting identifier, and / or delay from a measurement period of a channel state information reference signal.

[0084] Some exemplary embodiments may provide that the one or more delay requirements for the switching operation includes at least one of: delay from selecting a semi-persistent channel state information resource set, delay from processing by the apparatus of a capability report of the apparatus, delay from obtaining an associated reporting identifier, or delay from a measurement period of a channel state information reference signal. The one or more delay requirements for the fallback operation may include at least one of: delay for activating a semi-persistent channel state information resource set, delay for selecting the semi-persistent channel state information resource set, delay from processing by the apparatus of a capability report of the apparatus, and / or delay from a measurement period of a channel state information reference signal.

[0085] Certain exemplary embodiments may provide that the measurement period of the channel state information reference signal is a time period for measuring a power level of the channel state information reference signal. The one or more delay requirements may include a processing time for the apparatus to generate a report of the inference results. The processing time may include at least one of a delay from activating a semi-persistent channel state information resource set triggered by downlink control information, or a delay for selecting the semi-persistent channel state information resource set triggered by a control element for medium access control. A delay in the processing time may include at least one of a preparation time for the apparatus to perform beam identification prediction, or a preparation time for the apparatus to perform beam identification prediction and reference signal power prediction. The delay in the processing time may include a processing time for obtaining the one or more inference results of the AI / ML beam management for beam prediction. The delay in the processing time may include a buffering time for an AI / MLfunctionality of the AI / ML beam management. The buffering time may include an amount of time to load an AI / ML model and for pre and post processing of data processed by the AI / ML model, and / or an amount of time for the apparatus to receive, from the network entity, the semi-persistent channel state information resource set.

[0086] FIG. 5 illustrates an example flow diagram of a method, according to certain exemplary embodiments. In an example embodiment, the method of FIG. 5 may be performed by a network element / entity, or a group of multiple network entities in a 3GPP system, such as LTE, 5G-NR, or 6G. For instance, in an exemplary embodiment, the method of FIG.5 may be performed by a network node or network entity / apparatus, such as a gNB, similar to apparatus 820 illustrated in FIG.8.

[0087] According to various exemplary embodiments, the method of FIG.5 may include, at 510, determining a configuration for artificial intelligence / machine learning (AI / ML) beam management to be performed by a user device, and at 520, determining one or more delay requirements associated with the AI / ML beam management. The method may also include, at 530, determining a time instance based on the determined one or more delay requirements associated with the AI / ML beam management, and at 540, transmitting, to the user device, the configuration for AI / ML beam management. The configuration may indicate a reporting frequency for the AI / ML beam management as semi-persistent reporting. The method may further include, at 550, receiving, via semi-persistent reporting from the user device during the time instance, a report indicating one or more inference results of the AI / ML beam management.

[0088] Certain exemplary embodiments may provide that the configuration for AI / ML beam management also indicates a life cycle management operation comprising at least one of the following: an activation operation for AI / ML inference, a deactivation operation for AI / ML inference, a switching operation from a first AI / ML inference to a second AI / ML inference, or a fallback operation to change from an AI / ML inference to a legacy non-AI / ML inference. The one or more delay requirements for the activation operation or the deactivation operation may include at least one of: delay for activating a semi-persistent channel state information resource set, delay for selecting the semi-persistent channel state information resource set, delay from processing by the apparatus of a capability report of the apparatus, delay from obtaining an associated reporting identifier, and / or delay from a measurement period of a channel state information reference signal.

[0089] Some exemplary embodiments may provide that the one or more delay requirements for the switching operation comprises at least one of: delay from selecting a semi-persistent channel state information resource set, delay from processing by the apparatus of a capability report of the apparatus, delay from obtaining an associated reporting identifier, and / or delay from a measurement period of a channel state information reference signal. The one or more delay requirements for the fallback operation may include at least one of: delay for activating a semi-persistent channel state information resource set, delay for selecting the semi-persistent channel state information resource set, delay from processing by the apparatus of a capability report of the apparatus, and / or delay from a measurement period of a channel stateinformation reference signal.

[0090] Various exemplary embodiments may provide that the measurement period of the channel state information reference signal is a time period for measuring a power level of the channel state information reference signal. The one or more delay requirements may include a processing time for the apparatus to generate a report of the inference results. The processing time may include at least one of a delay from activating the semi-persistent channel state information resource set triggered by downlink control information, or a delay for selecting the semi-persistent channel state information resource set triggered by a control element for medium access control. A delay in the processing time may include at least one of a preparation time for the user device to perform beam identification prediction, or a preparation time for the user device to perform beam identification prediction and reference signal power prediction. The delay in the processing time may include a processing time for the user device to obtain the one or more inference results of the AI / ML beam management for beam prediction. The delay in the processing time may include a buffering time for an AI / ML functionality of the AI / ML beam management. The buffering time may include an amount of time for the user device to load an AI / ML model and for pre and post processing of data processed by the AI / ML model, and / or an amount of time for the apparatus to select and transmit, to the user device, the semi-persistent channel state information resource set.

[0091] FIG. 6 illustrates an example flow diagram of a method, according to certain exemplary embodiments. In an example embodiment, the method of FIG.6 may be performed by a user device in a 3GPP system, such as LTE, 5G-NR, or 6G. For instance, in an exemplary embodiment, the method of FIG. 6 may be performed by a user device / apparatus or user equipment, such as a UE, similar to apparatus 810 illustrated in FIG.8.

[0092] According to various exemplary embodiments, the method of FIG.6 may include, at 610, receiving, from a network entity, a configuration for artificial intelligence / machine learning (AI / ML) beam management. The configuration may indicate a reporting frequency for AI / ML beam management as aperiodic reporting. The method may also include, at 620, determining one or more delay requirements associated with the AI / ML beam management, and at 630, determining a time instance based on the determined one or more delay requirements associated with the AI / ML beam management. The method may further include, at 640, performing the AI / ML beam management, and at 650, transmitting, via aperiodic reporting to the network entity during the time instance, a report indicating one or more inference results of the AI / ML beam management.

[0093] Certain exemplary embodiments may provide the configuration for AI / ML beam management also indicates a life cycle management operation. The life cycle management operation may include at least one of the following: an activation operation for AI / ML inference, a deactivation operation for AI / ML inference, a switching operation from a first AI / ML inference to a second AI / ML inference, or a fallback operation to change from an AI / ML inference to a legacy non-AI / ML inference. The one or more delay requirements for theactivation operation or the deactivation operation may include at least one of: delay from processing by the apparatus of a capability report of the apparatus, delay from obtaining an associated reporting identifier, and / or delay from a measurement period of a channel state information reference signal.

[0094] Some exemplary embodiments may provide that the one or more delay requirements for the switching operation comprises at least one of: delay from processing by the apparatus of a capability report of the apparatus, delay from obtaining an associated reporting identifier, and / or delay from a measurement period of a channel state information reference signal. The measurement period of the channel state information reference signal may be a time period for measuring a power level of the channel state information reference signal. The one or more delay requirements may include a processing time for the apparatus to generate a report of the inference results. The processing time may include a delay from activating an aperiodic channel state information resource set triggered by downlink control information. A delay in the processing time may include at least one of a preparation time for the apparatus to perform beam identification prediction or a preparation time for the apparatus to perform beam identification prediction and reference signal power prediction. The delay in the processing time may include a processing time for obtaining the one or more inference results of the AI / ML beam management for beam prediction. The delay in the processing time may include a buffering time for an AI / ML functionality of the AI / ML beam management. The buffering time may include an amount of time to load an AI / ML model and for pre and post processing of data processed by the AI / ML model and / or an amount of time for the apparatus to receive, from the network entity, the aperiodic channel state information resource set.

[0095] FIG. 7 illustrates an example flow diagram of a method, according to certain exemplary embodiments. In an example embodiment, the method of FIG. 7 may be performed by a network element / entity, or a group of multiple network entities in a 3GPP system, such as LTE, 5G-NR, or 6G. For instance, in an exemplary embodiment, the method of FIG.7 may be performed by a network node or network entity / apparatus, such as a gNB, similar to apparatus 820 illustrated in FIG.8.

[0096] According to various exemplary embodiments, the method of FIG.7 may include, at 710, determining a configuration for artificial intelligence / machine learning (AI / ML) beam management to be performed by a user device, and at 720, determining one or more delay requirements associated with the AI / ML beam management. The method may also include, at 730, determining a time instance based on the determined one or more delay requirements associated with the AI / ML beam management, and at 740, transmitting, to the user device, the configuration for AI / ML beam management. The configuration may indicate a reporting frequency for the AI / ML beam management as aperiodic reporting. The method may further include, at 750, receiving, via aperiodic reporting from the user device during the time instance, a report indicating one or more inference results of the AI / ML beam management.

[0097] Certain exemplary embodiments may provide the configuration for AI / ML beam management also indicates a life cycle management operation. The life cycle management operation may include at least oneof the following: an activation operation for AI / ML inference, a deactivation operation for AI / ML inference, a switching operation from a first AI / ML inference to a second AI / ML inference, or a fallback operation to change from an AI / ML inference to a legacy non-AI / ML inference. The one or more delay requirements for the activation operation or the deactivation operation may include at least one of: delay from processing by the apparatus of a capability report of the apparatus, delay from obtaining an associated reporting identifier, an / or delay from a measurement period of a channel state information reference signal. The one or more delay requirements for the switching operation may include at least one of: delay from the user device performing a new radio resource control reconfiguration, which is different from a prior radio resource control configuration, delay from processing by the apparatus of a capability report of the apparatus, delay from obtaining an associated reporting identifier, and / or delay from a measurement period of a channel state information reference signal.

[0098] Some exemplary embodiments may provide that the measurement period of the channel state information reference signal is a time period for measuring a power level of the channel state information reference signal. The one or more delay requirements may include a processing time for the apparatus to generate a report of the inference results. The processing time may include a delay from activating an aperiodic channel state information resource set triggered by downlink control information. A delay in the processing time comprises at least one of: a preparation time for the user device to perform beam identification prediction, or a preparation time for the user device to perform beam identification prediction and reference signal power prediction. The delay in the processing time may include a processing time for the user device to obtain the one or more inference results of the AI / ML beam management for beam prediction. The delay in the processing time may include a buffering time for an AI / ML functionality of the AI / ML beam management. The buffering time may include an amount of time to load an AI / ML model and for pre and post processing of data processed by the AI / ML model, and / or an amount of time for the apparatus to receive, from the network entity, the aperiodic channel state information resource set.

[0099] FIG.8 illustrates apparatuses 810 and 820 according to various exemplary embodiments. In the various exemplary embodiments, the apparatus 810 may be an element in a network or associated with such a network, such as mobile device, user device, or other type of user equipment. UE 101 may be an example of apparatus 810 according to various exemplary embodiments as discussed above. It should be noted that one of ordinary skill in the art would understand that apparatus 810 may include components or features not shown in FIG.8. Further, apparatus 820 may be an element in a network or associated with such a network, such as a base station, gNB, and the like. gNB 102 may be an example of apparatus 820 according to various exemplary embodiments as discussed above. It should be noted that one of ordinary skill in the art would understand that apparatus 820 may include components or features not shown in FIG.8.

[0100] According to various exemplary embodiments, the apparatuses 810 and / or 820 may include one or more processors, one or more computer-readable storage medium (for example, memory, storage, or thelike), one or more radio access components (for example, a modem, a transceiver, or the like), and / or a user interface. In some exemplary embodiments, apparatuses 810 and / or 820 may be configured to operate using one or more radio access technologies, such as GSM, LTE, LTE-A, NR, 5G, WLAN, WiFi, NB-IoT, Bluetooth, NFC, MulteFire, and / or any other radio access technologies.

[0101] As illustrated in the example of FIG.8, apparatuses 810 and / or 820 may include or be coupled to processors 812 and 822, respectively, for processing information and executing instructions or operations. Processors 812 and 822 may be any type of general or specific purpose processor. In fact, processors 812 and 822 may include one or more of general-purpose computers, special purpose computers, microprocessors, digital signal processors (DSPs), field-programmable gate arrays (FPGAs), application- specific integrated circuits (ASICs), and processors based on a multi-core processor architecture, as examples. While a single processor 812 (822) for each of apparatuses 810 and / or 820 is shown in FIG.8, multiple processors may be utilized according to other example embodiments. For example, it should be understood that, in certain exemplary embodiments, apparatuses 810 and / or 820 may include two or more processors that may form a multiprocessor system (for example, in this case processors 812 and 822 may represent a multiprocessor) that may support multiprocessing. According to certain exemplary embodiments, the multiprocessor system may be tightly coupled or loosely coupled to, for example, form a computer cluster).

[0102] Processors 812 and 822 may perform functions associated with the operation of apparatuses 810 and / or 820, respectively, including, as some examples, precoding of antenna gain / phase parameters, encoding and decoding of individual bits forming a communication message, formatting of information, and overall control of the apparatuses 810 and / or 820, including processes illustrated in FIGs.1-7.

[0103] Apparatuses 810 and / or 820 may further include or be coupled to memory 814 and / or 824 (internal or external), respectively, which may be coupled to processors 812 and 822, respectively, for storing information and instructions that may be executed by processors 812 and 822. Memory 814 (memory 824) may be one or more memories and of any type suitable to the local application environment, and may be implemented using any suitable volatile or nonvolatile data storage technology such as a semiconductor- based memory device, a magnetic memory device and system, an optical memory device and system, fixed memory, and / or removable memory. For example, memory 814 (memory 824) can be comprised of any combination of random access memory (RAM), read only memory (ROM), static storage such as a magnetic or optical disk, hard disk drive (HDD), or any other type of non-transitory machine or computer readable media. The instructions stored in memory 814 and memory 824 may include program instructions or computer program code that, when executed by processors 812 and 822, enable the apparatuses 810 and / or 820 to perform tasks as described herein.

[0104] In certain exemplary embodiments, apparatuses 810 and / or 820 may further include or be coupled to (internal or external) a drive or port that is configured to accept and read an external computer readablestorage medium, such as an optical disc, USB drive, flash drive, or any other storage medium. For example, the external computer readable storage medium may store a computer program or software for execution by processors 812 and 822 and / or apparatuses 810 and / or 820 to perform any of the methods illustrated in FIGs.1-7.

[0105] In some exemplary embodiments, apparatuses 810 and / or 820 may also include or be coupled to one or more antennas 815 and 825, respectively, for receiving a downlink signal and for transmitting via an uplink from apparatuses 810 and / or 820. Apparatuses 810 and / or 820 may further include transceivers 816 and 826, respectively, configured to transmit and receive information. The transceivers 816 and 826 may also include a radio interface (for example, a modem) respectively coupled to the antennas 815 and 825. The radio interface may correspond to a plurality of radio access technologies including one or more of GSM, LTE, LTE-A, 5G, NR, WLAN, NB-IoT, Bluetooth, BT-LE, NFC, RFID, UWB, or the like. The radio interface may include other components, such as filters, converters (for example, digital-to-analog converters or the like), symbol demappers, signal shaping components, an Inverse Fast Fourier Transform (IFFT) module, or the like, to process symbols, such as OFDMA symbols, carried by a downlink or an uplink.

[0106] For instance, transceivers 816 and 826 may be respectively configured to modulate information on to a carrier waveform for transmission by the antenna(s) 815 and 825, and demodulate information received via the antenna(s) 815 and 825 for further processing by other elements of apparatuses 810 and / or 820. In other exemplary embodiments, transceivers 816 and 826 may be capable of transmitting and receiving signals or data directly. Additionally or alternatively, in some exemplary embodiments, apparatuses 810 and / or 820 may include an input and / or output device (I / O device). In certain exemplary embodiments, apparatuses 810 and / or 820 may further include a user interface, such as a graphical user interface or touchscreen.

[0107] In certain exemplary embodiments, memory 814 and memory 824 store software modules that provide functionality when executed by processors 812 and 822, respectively. The modules may include, for example, an operating system that provides operating system functionality for apparatuses 810 and / or 820. The memory may also store one or more functional modules, such as an application or program, to provide additional functionality for apparatuses 810 and / or 820. The components of apparatuses 810 and / or 820 may be implemented in hardware, or as any suitable combination of hardware and software. According to certain exemplary embodiments, apparatus 810 may optionally be configured to communicate with apparatus 820 via a wireless or wired communications link 830 according to any radio access technology, such as NR.

[0108] According to certain exemplary embodiments, processors 812 and 822, and memory 814 and 824 may be included in or may form a part of processing circuitry or control circuitry. In addition, in some exemplary embodiments, transceivers 816 and 826 may be included in or may form a part of transceiving circuitry.

[0109] For instance, in certain exemplary embodiments, the apparatus 810 may be controlled by thememory 814 and the processor 812 to receive, from a network entity, a configuration for artificial intelligence / machine learning (AI / ML) beam management. The configuration may indicate a reporting frequency for the AI / ML beam management as periodic reporting. The apparatus 810 may also be controlled to determine one or more delay requirements associated with the AI / ML beam management and determine a time instance based on the determined one or more delay requirements associated with the AI / ML beam management. The apparatus 810 may further be controlled to perform the AI / ML beam management and transmit, via periodic reporting to the network entity during the time instance, a report indicating one or more inference results of the AI / ML beam management.

[0110] In various exemplary embodiments, the apparatus 820 may be controlled by the memory 824 and the processor 822 to determine a configuration for artificial intelligence / machine learning (AI / ML) beam management to be performed by a user device and determine one or more delay requirements associated with the AI / ML beam management. The apparatus 820 may also be controlled to determine a time instance based on the determined one or more delay requirements associated with the AI / ML beam management and transmit to the user device, the configuration for AI / ML beam management. The configuration may indicate a reporting frequency for the AI / ML beam management as periodic reporting. The apparatus 820 may further be controlled to receive, via periodic reporting from the user device during the time instance, a report indicating one or more inference results of the AI / ML beam management.

[0111] In some additional exemplary embodiments, the apparatus 810 may be controlled by the memory 814 and the processor 812 to receive, from a network entity, a configuration for artificial intelligence / machine learning (AI / ML) beam management. The configuration may indicate a reporting frequency for the AI / ML beam management as semi-persistent reporting. The apparatus 810 may also be controlled to determine one or more delay requirements associated with the AI / ML beam management, and determine a time instance based on the determined one or more delay requirements associated with the AI / ML beam management. The apparatus 810 may further be controlled to perform the AI / ML beam management and transmit, via semi-persistent reporting to the network entity during the time instance, a report indicating one or more inference results of the AI / ML beam management.

[0112] In certain additional exemplary embodiments, the apparatus 820 may be controlled by the memory 824 and the processor 822 to determine a configuration for artificial intelligence / machine learning (AI / ML) beam management to be performed by a user device and determine one or more delay requirements associated with the AI / ML beam management. The apparatus 820 may also be controlled to determine a time instance based on the determined one or more delay requirements associated with the AI / ML beam management and transmit, to the user device, the configuration for AI / ML beam management. The configuration may indicate a reporting frequency for the AI / ML beam management as semi-persistent reporting. The apparatus 820 may further be controlled to receive, via semi-persistent reporting from the user device during the time instance, a report indicating one or more inference results of the AI / ML beammanagement.

[0113] In various other exemplary embodiments, the apparatus 810 may be controlled by the memory 814 and the processor 812 to receive, from a network entity, a configuration for artificial intelligence / machine learning (AI / ML) beam management. The configuration may indicate a reporting frequency for AI / ML beam management as aperiodic reporting. The apparatus 810 may also be controlled to determine one or more delay requirements associated with the AI / ML beam management, determine a time instance based on the determined one or more delay requirements associated with the AI / ML beam management, perform the AI / ML beam management, and transmit, via aperiodic reporting to the network entity during the time instance, a report indicating one or more inference results of the AI / ML beam management.

[0114] In certain other exemplary embodiments, the apparatus 820 may be controlled by the memory 824 and the processor 822 to determine a configuration for artificial intelligence / machine learning (AI / ML) beam management to be performed by a user device and determine one or more delay requirements associated with the AI / ML beam management. The apparatus 820 may also be controlled to determine a time instance based on the determined one or more delay requirements associated with the AI / ML beam management and transmit, to the user device, the configuration for AI / ML beam management. The configuration may indicate a reporting frequency for the AI / ML beam management as aperiodic reporting. The apparatus 820 may further be controlled to receive, via aperiodic reporting from the user device during the time instance, a report indicating one or more inference results of the AI / ML beam management.

[0115] In some exemplary embodiments, an apparatus (e.g., apparatus 810 and / or apparatus 820) may include means for performing a method, a process, or any of the variants discussed herein. Examples of the means may include one or more processors, memory, controllers, transmitters, receivers, and / or computer program code for causing the performance of the operations.

[0116] In various exemplary embodiments, the apparatus 810 may include means for receiving, from a network entity, a configuration for artificial intelligence / machine learning (AI / ML) beam management. The configuration may indicate a reporting frequency for the AI / ML beam management as periodic reporting. The apparatus 810 may also include means for determining one or more delay requirements associated with the AI / ML beam management and means for determining a time instance based on the determined one or more delay requirements associated with the AI / ML beam management. The apparatus 810 may further include means for performing the AI / ML beam management and means for transmitting, via periodic reporting to the network entity during the time instance, a report indicating one or more inference results of the AI / ML beam management.

[0117] In some exemplary embodiments, the apparatus 820 may include means for determining a configuration for artificial intelligence / machine learning (AI / ML) beam management to be performed by a user device and means for determining one or more delay requirements associated with the AI / ML beam management. The apparatus 820 may also include means for determining a time instance based on thedetermined one or more delay requirements associated with the AI / ML beam management and means for transmitting to the user device, the configuration for AI / ML beam management. The configuration may indicate a reporting frequency for the AI / ML beam management as periodic reporting. The apparatus 820 may further include means for receiving, via periodic reporting from the user device during the time instance, a report indicating one or more inference results of the AI / ML beam management.

[0118] In certain additional exemplary embodiments, the apparatus 810 may include means for receiving, from a network entity, a configuration for artificial intelligence / machine learning (AI / ML) beam management. The configuration may indicate a reporting frequency for the AI / ML beam management as semi-persistent reporting. The apparatus 810 may also include means for determining one or more delay requirements associated with the AI / ML beam management, and means for determining a time instance based on the determined one or more delay requirements associated with the AI / ML beam management. The apparatus 810 may further include means for performing the AI / ML beam management and means for transmitting, via semi-persistent reporting to the network entity during the time instance, a report indicating one or more inference results of the AI / ML beam management.

[0119] In some additional exemplary embodiments, the apparatus 820 may include means for determining a configuration for artificial intelligence / machine learning (AI / ML) beam management to be performed by a user device and means for determining one or more delay requirements associated with the AI / ML beam management. The apparatus 820 may also include means for determining a time instance based on the determined one or more delay requirements associated with the AI / ML beam management and means for transmitting, to the user device, the configuration for AI / ML beam management. The configuration may indicate a reporting frequency for the AI / ML beam management as semi-persistent reporting. The apparatus 820 may further include means for receiving, via semi-persistent reporting from the user device during the time instance, a report indicating one or more inference results of the AI / ML beam management.

[0120] In various other exemplary embodiments, the apparatus 810 may include means for receiving, from a network entity, a configuration for artificial intelligence / machine learning (AI / ML) beam management. The configuration may indicate a reporting frequency for AI / ML beam management as aperiodic reporting. The apparatus 810 may also include means for determining one or more delay requirements associated with the AI / ML beam management and means for determining a time instance based on the determined one or more delay requirements associated with the AI / ML beam management. The apparatus 810 may further include means for performing the AI / ML beam management and means for transmitting, via aperiodic reporting to the network entity during the time instance, a report indicating one or more inference results of the AI / ML beam management.

[0121] In certain other exemplary embodiments, the apparatus 820 may include means for determining a configuration for artificial intelligence / machine learning (AI / ML) beam management to be performed by a user device and means for determining one or more delay requirements associated with the AI / ML beammanagement. The apparatus 820 may also include means for determining a time instance based on the determined one or more delay requirements associated with the AI / ML beam management and means for transmitting, to the user device, the configuration for AI / ML beam management. The configuration may indicate a reporting frequency for the AI / ML beam management as aperiodic reporting. The apparatus 820 may further include means for receiving, via aperiodic reporting from the user device during the time instance, a report indicating one or more inference results of the AI / ML beam management.

[0122] As used herein, the term “circuitry” may refer to hardware-only circuitry implementations (for example, analog and / or digital circuitry), combinations of hardware circuits and software, combinations of analog and / or digital hardware circuits with software / firmware, any portions of hardware processor(s) with software, including digital signal processors, that work together to cause an apparatus (for example, apparatus 810 and / or 820) to perform various functions, and / or hardware circuit(s) and / or processor(s), or portions thereof, that use software for operation but where the software may not be present when it is not needed for operation. As a further example, as used herein, the term “circuitry” may also cover an implementation of merely a hardware circuit or processor or multiple processors, or portion of a hardware circuit or processor, and the accompanying software and / or firmware. The term circuitry may also cover, for example, a baseband integrated circuit in a server, cellular network node or device, or other computing or network device.

[0123] A computer program product may include one or more computer-executable components which, when the program is run, are configured to carry out some exemplary embodiments. The one or more computer-executable components may be at least one software code or portions of it. Modifications and configurations required for implementing functionality of certain exemplary embodiments may be performed as routine(s), which may be implemented as added or updated software routine(s). Software routine(s) may be downloaded into the apparatus.

[0124] As an example, software or a computer program code or portions of it may be in a source code form, object code form, or in some intermediate form, and it may be stored in some sort of carrier, distribution medium, or computer readable medium, which may be any entity or device capable of carrying the program. Such carriers may include a record medium, computer memory, read-only memory, photoelectrical and / or electrical carrier signal, telecommunications signal, and software distribution package, for example. Depending on the processing power needed, the computer program may be executed in a single electronic digital computer or it may be distributed amongst a number of computers. The computer readable medium or computer readable storage medium may be a non-transitory medium.

[0125] In other exemplary embodiments, the functionality may be performed by hardware or circuitry included in an apparatus (for example, apparatuses 810 and / or 820), for example through the use of an application specific integrated circuit (ASIC), a programmable gate array (PGA), a field programmable gate array (FPGA), or any other combination of hardware and software. In yet another exemplary embodiment,the functionality may be implemented as a signal, a non-tangible means that can be carried by an electromagnetic signal downloaded from the Internet or other network.

[0126] According to certain exemplary embodiments, an apparatus, such as a node, device, or a corresponding component, may be configured as circuitry, a computer or a microprocessor, such as single- chip computer element, or as a chipset, including at least a memory for providing storage capacity used for arithmetic operation and an operation processor for executing the arithmetic operation.

[0127] The features, structures, or characteristics of exemplary embodiments described throughout this specification may be combined in any suitable manner in one or more exemplary embodiments. For example, the usage of the phrases “certain embodiments,” “an example embodiment,” “some embodiments,” or other similar language, throughout this specification refers to the fact that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment. Thus, appearances of the phrases “in certain embodiments,” “an example embodiment,” “in some embodiments,” “in other embodiments,” or other similar language, throughout this specification do not necessarily refer to the same group of embodiments, and the described features, structures, or characteristics may be combined in any suitable manner in one or more exemplary embodiments. Further, the terms “cell”, “node”, “gNB”, or other similar language throughout this specification may be used interchangeably.

[0128] 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.

[0129] One having ordinary skill in the art will readily understand that the disclosure as discussed above may be practiced with procedures in a different order, and / or with hardware elements in configurations which are different than those which are disclosed. Therefore, although the disclosure has been described based upon these exemplary embodiments, it would be apparent to those of skill in the art that certain modifications, variations, and alternative constructions would be apparent, while remaining within the spirit and scope of exemplary embodiments. Although the above embodiments refer to 5G NR and LTE technology, the above embodiments may also apply to any other present or future 3GPP technology, such as LTE-advanced, and / or fourth generation (4G) and / or sixth (6G) technology.

[0130] Partial Glossary:

[0131] 3GPP 3rd Generation Partnership Project

[0132] 5G 5th Generation

[0133] AI Artificial Intelligence

[0134] CSI Channel State Information

[0135] CSI-RS Channel State Information Reference Signal

[0136] DCI Downlink Control Information

[0137] DL Downlink

[0138] EMBB Enhanced Mobile Broadband

[0139] gNB 5G or Next Generation NodeB

[0140] LCM Life Cycle Management

[0141] LTE Long Term Evolution

[0142] MAC-CE Medium Access Control Control Element

[0143] ML Machine Learning

[0144] NR New Radio

[0145] RF Radio Frequency

[0146] RRC Radio Resource Control

[0147] Tx Transmit

[0148] UE User Equipment

[0149] UL Uplink

Claims

WE CLAIM:

1. An apparatus comprising: at least one processor; and at least one memory storing instructions that, when executed by the at least one processor, cause the apparatus at least to: receive, from a network entity, a configuration for artificial intelligence / machine learning (AI / ML) beam management, wherein the configuration indicates a reporting frequency for AI / ML beam management as aperiodic reporting; determine one or more delay requirements associated with the AI / ML beam management; determine a time instance based on the determined one or more delay requirements associated with the AI / ML beam management; perform the AI / ML beam management; and transmit, via aperiodic reporting to the network entity during the time instance, a report indicating one or more inference results of the AI / ML beam management.

2. The apparatus according to claim 1, wherein the configuration for AI / ML beam management also indicates a life cycle management operation.

3. The apparatus according to claim 2, wherein the life cycle management operation comprises at least one of the following: an activation operation for AI / ML inference, a deactivation operation for AI / ML inference, a switching operation from a first AI / ML inference to a second AI / ML inference, or a fallback operation to change from an AI / ML inference to a legacy non-AI / ML inference.

4. The apparatus according to claim 3, wherein the one or more delay requirements for the activation operation or the deactivation operation comprises at least one of: delay from processing by the apparatus of a capability report of the apparatus; delay from obtaining an associated reporting identifier; or delay from a measurement period of a channel state information reference signal.

5. The apparatus according to claim 3, wherein the one or more delay requirements for the switching operation comprises at least one of: delay from processing by the apparatus of a capability report of the apparatus; delay from obtaining an associated reporting identifier; or delay from a measurement period of a channel state information reference signal.

6. The apparatus according to claim 4 or claim 5, wherein the measurement period of the channel state information reference signal is a time period for measuring a power level of the channel state information reference signal.

7. The apparatus according to any one of claims 1-6, wherein the one or more delay requirements comprises a processing time for the apparatus to generate a report of the inference results, wherein the processing time comprises a delay from activating an aperiodic channel state information resource set triggered by downlink control information.

8. The apparatus according to claim 7, wherein a delay in the processing time comprises at least one of: a preparation time for the apparatus to perform beam identification prediction, or a preparation time for the apparatus to perform beam identification prediction and reference signal power prediction.

9. The apparatus according to claim 7 or claim 8, wherein the delay in the processing time comprises a processing time for obtaining the one or more inference results of the AI / ML beam management for beam prediction.

10. The apparatus according to any one of claims 7-9, wherein the delay in the processing time comprises a buffering time for an AI / ML functionality of the AI / ML beam management, wherein the buffering time comprises: an amount of time to load an AI / ML model and for pre and post processing of data processed by the AI / ML model, and an amount of time for the apparatus to receive, from the network entity, the aperiodic channel state information resource set.

11. An apparatus comprising: at least one processor; and at least one memory storing instructions that, when executed by the at least one processor, cause the apparatus at least to: determine a configuration for artificial intelligence / machine learning (AI / ML) beam management to be performed by a user device; determine one or more delay requirements associated with the AI / ML beam management;determine a time instance based on the determined one or more delay requirements associated with the AI / ML beam management; transmit, to the user device, the configuration for AI / ML beam management, wherein the configuration indicates a reporting frequency for the AI / ML beam management as aperiodic reporting; and receive, via aperiodic reporting from the user device during the time instance, a report indicating one or more inference results of the AI / ML beam management.

12. The apparatus according to claim 11, wherein the configuration for AI / ML beam management also indicates a life cycle management operation.

13. The apparatus according to claim 12, wherein the life cycle management operation comprises at least one of the following: an activation operation for AI / ML inference, a deactivation operation for AI / ML inference, a switching operation from a first AI / ML inference to a second AI / ML inference, or a fallback operation to change from an AI / ML inference to a legacy non-AI / ML inference.

14. The apparatus according to claim 13, wherein the one or more delay requirements for the activation operation or the deactivation operation comprises at least one of: delay from processing by the apparatus of a capability report of the apparatus; delay from obtaining an associated reporting identifier; or delay from a measurement period of a channel state information reference signal.

15. The apparatus according to claim 13, wherein the one or more delay requirements for the switching operation comprises at least one of: delay from the user device performing a new radio resource control reconfiguration, which is different from a prior radio resource control configuration; delay from processing by the apparatus of a capability report of the apparatus; delay from obtaining an associated reporting identifier; or delay from a measurement period of a channel state information reference signal.

16. The apparatus according to claim 14 or claim 15, wherein the measurement period of the channel state information reference signal is a time period for measuring a power level of the channel state information reference signal.

17. The apparatus according to any one of claims 11-16, wherein the one or more delay requirements comprises a processing time for the apparatus to generate a report of the inference results, wherein theprocessing time comprises a delay from activating an aperiodic channel state information resource set triggered by downlink control information.

18. The apparatus according to claim 17, wherein a delay in the processing time comprises at least one of: a preparation time for the user device to perform beam identification prediction, or a preparation time for the user device to perform beam identification prediction and reference signal power prediction.

19. The apparatus according to claim 17 or claim 18, wherein the delay in the processing time comprises a processing time for the user device to obtain the one or more inference results of the AI / ML beam management for beam prediction.

20. The apparatus according to any one of claims 17-19, wherein the delay in the processing time comprises a buffering time for an AI / ML functionality of the AI / ML beam management, wherein the buffering time comprises: an amount of time to load an AI / ML model and for pre and post processing of data processed by the AI / ML model, and an amount of time for the apparatus to receive, from the network entity, the aperiodic channel state information resource set.

Images