WO2025207129A1

WO2025207129A1 - Interworking between ai-based and non-ai-based channel measurement reporting

Info

Publication number: WO2025207129A1
Application number: PCT/US2024/032354
Authority: WO
Inventors: Subramanya CHANDRASHEKAR; Raghavendra Madanahally RAMAKRISHNA
Original assignee: Rakuten Symphony Inc; Rakuten Mobile USA LLC
Current assignee: Rakuten Symphony Inc; Rakuten Mobile USA LLC
Priority date: 2024-03-27
Filing date: 2024-06-04
Publication date: 2025-10-02
Anticipated expiration: 2026-09-27

Abstract

A method is disclosed. The method includes receiving, by a gNB, one or more UE-based Artificial Intelligence/Machine Learning (AI/ML) model output parameters and/or Key Performance Indicators (KPIs) and device capability information from a User Equipment (UE). The method also includes comparing, by the gNB, the one or more UE-based AI/ML model output parameters and/or KPIs with corresponding predefined benchmarked values. Furthermore, the method includes upon determining that at least one of the one or more UE-based AI/ML model output parameters and/or KPIs is below the corresponding predefined benchmarked value configuring, by the gNB, the UE to perform non-AI/ML-based operations for the at least one selected functionality or use-case, and one of disabling or continuing AI/ML-based measurements and/or predictions for the at least one selected functionality or use-case based on one or more predefined criteria corresponding to the received device capability information.

Description

INTERWORKING BETWEEN AI-BASED AND NON-AI-BASED CHANNEL

MEASUREMENT REPORTING

CROSS-REFERENCE TO RELATED APPLICATION(S)

[001] This application claims the benefit of Indian Non-Provisional Application No 202411024625, entitled ‘INTERWORKING BETWEEN AI-BASED AND NON-AI-BASED CHANNEL MEASUREMENT REPORTING” and filed on March 27, 2024, which is expressly incorporated by reference herein in its entirety.

FIELD

[002] The present disclosure relates to an interworking between an Artificial Intelligence/Machine Learning (AI/ML) based and a non-AI/ML based channel measurement reporting in a wireless communication system.

BACKGROUND

[003] In wireless communication systems, particularly those employing multiple antennas at both a transmitter end and a receiver end, Channel State Information (CSI) plays a pivotal role in optimizing system performance. The CSI encompasses infomiation about characteristics of a wireless channel, including fading, attenuation, and noise properties. The information about the characteristics of the wireless channel enables implementation of advanced signal processing techniques such as beamforming, spatial multiplexing, and interference mitigation.

[004] In particular, a CSI feedback mechanism in the wireless communication systems involves transmission of information from a receiver back to a transmitter regarding a cunent state of the wireless communication channel as a CSI feedback. The CSI feedback allows the transmitter to adapt transmission parameters and strategies based on real-time channel conditions, optimizing communication reliability and spectral efficiency. The CSI feedback includes metrics such as channel magnitude, phase, coherence bandwidth, and Doppler spread. The CSI feedback enables the transmitter to employ advanced signal processing techniques like beamforming and spatial multiplexing. Such an adaptive approach improves system performance by mitigating the effects of channel fading, interference, and noise, thereby enhancing the overall qualify of service for users of the wireless communication systems across various applications and environments.

SUMMARY

[005] The summan' is provided to introduce a selection of concepts, in a simplified format, that are further described in the detailed description of the disclosure. The summary is neither intended to identify key or essential inventive concepts of the present disclosure nor is it intended to determine the scope of the disclosure.

[006] The integration of Machine Learning (ML) or Artificial Intelligence (Al)-driven approaches in CS1 feedback mechanisms, has impacted conventional non-Al/ML-based CS1 feedback mechanisms. In particular, a User Equipment (UE) may be configured to report the non-AI/ML and the AI/ML-based CSI feedback based on non-AI/ML-based or AI/ML-based CSI prediction mode being enabled or configured. However, when the AI/ML-based CSI feedback does not result in a desired performance (i.e.. Key Performance Indicators are negatively impacted), an overhead of the AI/ML-based CSI feedback should be avoided or minimized. Similarly, when there is a periodic event collision (or there is a race condition), a network should know which report has been sent by the UE. The periodic event collision or the race condition may correspond to a situation when the UE can send only one report, i.e., either the non-AI/ML-based CSI feedback or the AI/ML-based CSI feedback. In an absence of such awareness of the prediction mode, the network may not be able to take an appropriate decision on which prediction mode is to be enabled or disabled.

[007] An object of the present disclosure is to provide a technique interworking between AI/ML-based and non-AI/ML-based CSI feedback.

[008] According to one embodiment of the present disclosure, a method is disclosed. The method includes receiving, by a gNB, one or more UE-based Artificial Intelligence/Machine Learning (AI/ML) model output parameters and/or Key Performance Indicators (KPIs). The method also includes receiving device capability information. The one or more UE-based AI/ML model output parameters and/or KPIs and device capability information are received from a User Equipment (UE) for at least one selected functionality or use-case. Further, the method includes comparing, by the gNB, the one or more UE-based AI/ML model output parameters and/or KPIs with corresponding predefined benchmarked values. The method further includes, upon determining that at least one of the one or more AI/ML model output parameters and/or KPIs is below the corresponding predefined benchmarked value configuring, by the gNB, the UE. The method includes configuring the UE to fallback to non-AI/ML-based operations for the at least one selected functionality or use-case. Furthermore, the method includes configuring the UE to one of disable or continue AI/ML-based measurements and/or predictions for the at least one selected functionality or use-case. The UE is configured to one of disable or continue AI/ML-based measurements and/or predictions based on one or more predefined criteria corresponding to the received device capability' information.

[009] According to another embodiment of the present disclosure, an apparatus is disclosed. The apparatus is configured to receive one or more UE-based Artificial Intelligence/Machine Learning (AI/ML) model output parameters and/or Key Performance Indicators (KPIs). The apparatus is also configured to receive device capability information. The one or more UE- based AI/ML model output parameters and/or KPIs and the device capability information is received from a user equipment (UE) for at least one selected functionality or use-case. The apparatus is also configured to compare the one or more UE-based AI/ML model output parameters and/or KPIs with corresponding predefined benchmarked values. Upon determining that at least one of the one or more UE-based AI/ML model output parameters and/or KPIs is below the corresponding predefined benchmarked value, the apparatus is configured to configure the UE. The apparatus is configured to configure the UE to fallback to non-AI/ML- based operations for the at least one selected functionality or use-case. The apparatus is also configured to configure the UE to one of disable or continue AI/ML-based measurements and/or predictions for the at least one selected functionality or use-case based on one or more predefined criteria corresponding to the received device capability information.

[0010] According to another embodiment of the present disclosure, a non-transitory computer- readable medium storing instructions is disclosed. The instructions include one or more instructions that are executed by a network device comprising one or more processors. The instructions cause the one or more processors to receive one or more UE-based Artificial Intelligence/Machine Learning (AI/ML) model output parameters and/or Key Perfonnance Indicators (KPIs) and device capability information. The one or more UE-based AI/ML model output parameters and/or KPIs and the device capability information are received from a user equipment (UE) for at least one selected functionality or use-case. The instructions cause the one or more processors to compare the one or more UE-based AI/ML model output parameters and/or KPIs with corresponding predefined benchmarked values. Upon determining that at least one of the one or more UE-based AI/ML model output parameters and/or KPIs is below the corresponding predefined benchmarked value, the instructions cause the one or more processors to configure the UE. The UE is configured to fallback to non-AI/ML-based operations for the at least one selected functionality or use-case. The UE is also configured to one of disable or continue AI/ML-based measurements and/or predictions for the at least one selected functionality or use-case based on one or more predefined criteria corresponding to the received device capability information.

[0011] To further clarity the advantages and features of the present disclosure, a more particular description of the disclosure will be rendered by reference to specific embodiments thereof, which is illustrated in the appended draw ing. It is appreciated that these drawings depict only typical embodiments of the disclosure and are therefore not to be considered limiting its scope. The disclosure will be described and explained with additional specificity and detail with the accompanying drawings.

BRIEF DESCRIPTION OF DRAWINGS

[0012] Features, aspects, and advantages of certain embodiments of the disclosure will be described below with reference to the accompanying drawings, in which like reference numerals denote like elements, and wherein:

FIG. 1 illustrates an environment of a wireless network, according to various embodiments of the present disclosure; FIG. 2 illustrates a functional framework for an AI/ML model for a New Radio (NR) air interface, according to various embodiments of the present disclosure.

FIG. 3 illustrates a sequence of operations between a User Equipment (UE) and a network entity (gNB), according to various embodiments of the present disclosure;

FIG. 4 illustrates another sequence of operations between the UE and the network entity (gNB), according to various embodiments of the present disclosure;

FIG. 5 illustrates a block diagram of the gNB, according to various embodiments of the present disclosure;

FIGS. 6a-6c illustrate a flow chart of an example method, in accordance with an embodiment of the present disclosure;

FIG. 7 illustrates a flow chart of an example method, in accordance with another embodiment of the present disclosure; and

FIG. 8 is a diagram of example components of a device of the wireless network, according to various embodiments of the present disclosure.

DETAILED DESCRIPTION

[0013] The following detailed description of example embodiments refers to the accompanying drawings. The foregoing disclosure provides illustration and description, but is not intended to be exhaustive or to limit the implementations to the precise form disclosed. Modifications and variations are possible in light of the above disclosure or may be acquired from practice of the implementations. Further, one or more features or components of one embodiment may be incorporated into or combined with another embodiment (or one or more features of another embodiment). Additionally, in the flowcharts and descriptions of operations provided below, it is understood that one or more operations may be omitted, one or more operations may be added, one or more operations may be performed simultaneously (at least in part), and the order of one or more operations may be switched, as long as these modifications may not affect the resulting scope of the disclosure.

[0014] It will be apparent that systems and/or methods, described herein, may be implemented in different forms of hardware, software, or a combination of hardware and software. The actual specialized control hardware or software code used to implement these systems and/or methods is not limiting the implementations. Thus, the operation and behavior of the systems and/or methods were described herein without reference to specific software code. It is understood that software and hardware may be designed to implement the systems and/or methods based on the description herein.

[0015] Even though particular combinations of features are recited in the claims and/or disclosed in the specification, these combinations are not intended to limit the disclosure of possible implementations. In fact, many of these features may be combined in ways not specifically recited in the claims and/or disclosed in the specification. Although each dependent claim listed below may directly depend on only one claim, the disclosure of possible implementations includes each dependent claim in combination with every other claim in the claim set.

[0016] No element, act, or instruction used herein should be construed as critical or essential unless explicitly described as such. Also, as used herein, the articles ‘’a” and “an” are intended to include one or more items, and may be used interchangeably with “one or more.” Where only one item is intended, the term “one” or similar language is used. Also, as used herein, the terms “has,” “have,” “having,” “include,” “including,” or the like are intended to be open-ended terms. Further, the phrase “based on” is intended to mean “based, at least in part, on” unless explicitly stated otherwise. Furthermore, expressions such as “at least one of [A] and [B]”, “[A] and/or [B]”, or “at least one of [A] or [B]” are to be understood as including only A, only B, or both A and B.

[0017] The foregoing disclosure provides illustration and description but is not intended to be exhaustive or to limit the implementations to the precise form disclosed. Modifications and variations are possible in light of the above disclosure or may be acquired from the practice of the implementations. [0018] The present disclosure relates to a technique to support interworking between an Artificial Intelligence/Machine Learning (AI/ML) -based and a non-AI/ML-based operations effectively and efficiently. The Channel State Information (CSI) feedback management usecase is considered as an example. With recent developments in wireless communication systems, the implementation of AI/ML-based models has been explored. In particular, to collect data and enhance performance for the New Radio (NR), the AI/ML-based measurement techniques have been provided that improve measurement parameters such as throughput, robustness, accuracy, reliability, etc. Such measurements correspond to CSI feedback, beam management, positioning accuracy enhancements, and so forth use-cases. The AI/ML-based models correspond to a data-driven approach that implements AI/ML techniques to generate a set of outputs based on an input dataset. Said AI/ML-based models are trained in a data-driven manner to obtain desired results.

[0019] In some embodiments, the AI/ML models are implemented in a pair of models configured to generate a joint inference of the desired result. For instance, in a wireless network, one of the AI/ML models is implemented at a User Equipment (UE) side and another AI/ML model is implemented at a network entity (for example, a base station). In such a scenario, a part of the joint inference is generated by one of the AI/ML models, and the remaining part of the inference is generated by another AI/ML model. For instance, in the wireless network, one portion of the inference is generated by the AI/ML model implemented at the UE, and the remaining portion of the inference is generated by the AI/ML model implemented at the network entity.

[0020] The AI/ML models may be analysed and updated based on associated model output parameters. The model output parameters may include one or more Key Performance Indicators (KPIs) associated with such AI/ML models. Further, said AI/ML models may be implemented via a management function that is configured to perfomi one or more of selecting an AI/ML model, deselecting the AI/ML model, activating the AI/ML model, deactivating the AI/ML model, and fallback to non-AI/ML based operations. However, it is required to effectively implement such a management function to reduce overhead and avoid a RACE condition. The RACE condition may correspond to a situation when the UE may be able to transmit one kind of prediction i.e.. either AI/ML-based prediction or non-AI/ML-based operations. [0021] In particular, the present disclosure provides analyzing the one or more KPIs associated with the AI/ML model and device capability information to determine when to configure the UE to perform non-AI/ML-based operations for the one or more selected functionality or usecase. The present disclosure also includes utilizing the analysed infonnation to determine whether to disable or continue AI/ML-based predictions for one or more selected functionality or use-cases. Thus, the present disclosure is able to effectively reduce the overhead and enable the AI/ML-based prediction to improve throughput, robustness, accuracy, or reliability, associated with the one or more selected functionalities or use-cases.

[0022] The tenns “base station’⁷, “gNB”, and “apparatus” have been used interchangeably throughout the description.

[0023] FIG. 1 illustrates an environment 100 of a wireless network in which systems and/or methods, described herein, may be implemented. The environment 100 may include a plurality of User Equipment (UEs) 102a-102n (interchangeably referred to as the UE 102 hereinafter) in communication with a base station (gNB) 106 via a communication network 104. The environment 100 as illustrated in FIG. 1 either covers or intends to cover any other suitable embodiment of the environment 100 without departing the scope of disclosure.

[0024] According to one or more embodiments of the present disclosure, the UE 102 may correspond to any remote wireless device that wirelessly communicates with the base station 106. An example of the UE 102 may include, but is not limited to, a mobile phone, a laptop computer, a desktop computer, a Personal Computer (PC), a notebook, a tablet, or any other suitable communication device having a capability to communicate with the base station 106 via the communication netw ork 104. In some other embodiments, the UE 102 may also be referred to as other well-known terms, such as a Mobile Station (MS), a Subscription Station (SS), a Remote Terminal (RT), a Wireless Terminal (WT), and the like.

[0025] In general, the UE 102 may communicate with the base station 106 using a plurality of uplink and downlink channels. Said plurality’ of uplink and downlink channels provide the means for transmitting data, voice, and other communication signals between the UE 102 and the base station 106. [0026] The base station 106 may be a computer system that is equivalent to an eNodeB (eNB) in a Fourth Generation (4G) network or a New Radio (NR) base station (gNB) in a Fifth Generation (5G) network. The gNB 106 according to the present embodiment may include one or more servers. The gNB 106 may be implemented by server groups disposed at data centers.

[0027] In an embodiment, the gNB 106 may be configured to receive one or more UE-based Artificial Intelligence/Machine Learning (AI/ML) model output parameters and/or KPIs and device capability' information from UE 102 for a selected functionality' or a use-case. In an example, the gNB 106 may be configured to receive the AI/ML model output parameters and/or KPIs at a trigger event. The trigger event may refer to an initiating event for receiving the AI/ML model output parameters and/or KPIs by the gNB 106 from the UE 102. In an aspect of the present disclosure, the trigger event may provide flexibility⁷ and adaptability' to the operations of the gNB 106.

[0028] In an example, the trigger event (also referred to as "periodic event⁷’), may correspond to a scheduled interval at which the gNB 106 may be configured to collect the AI/ML output parameters and/or KPIs from the UE 102. For example, the gNB 106 may request updates from the UE 102 every' few seconds, minutes, or hours, depending on the operational requirements and network conditions. Thus, the periodic event may ensure that the gNB 106 may receive consistent and regular updates on the AI/ML output parameters and/or KPIs.

[0029] Further, in an example, the gNB 106 may request specific information from the UE 102 outside of the periodic event. For instance, when the network 104 may experience sudden changes or when the gNB 106 may require immediate updates to address performance issues or optimize resource allocation. Accordingly, the gNB 106 may trigger a request to the UE 102, thus, prompting the UE 102 to send the AI/ML output parameters and/or KPIs.

[0030] Furthermore, in an example, the trigger event may include the UE 102 autonomously sending the AI/ML output parameters and/or KPIs to the gNB 106 based on certain conditions or events, as detected by the UE 102. For instance, if the UE 102 detects a significant change in the environment or operating conditions that could impact network performance, the UE 102 may initiate the transmission of the AI/ML output parameters and/or KPIs to the gNB 106 without waiting for a request.

[0031] In an embodiment, the gNB 106 may be configured to receive the AI/ML model output parameters and/or KPIs using a Layer 2 (L2) or a Layer 3 (L3) of the communication network 104. In an example, the L2 may also be referred to as a data link layer. The L2 may handle operations such as, but not limited to, framing, addressing, and error detection of data packets as transmitted between adjacent network nodes. L2 messages may encapsulate data packets and include information necessary for the delivery across physical links. Further, in an example, L3 may correspond to a network layer. The L3 may be configured for routing and forwarding data packets between different networks. The L3 may manage logical addressing, such as Internet Protocol (IP) addresses, and determine an optimal path for a packet delivery based on the network topology' and the routing algorithms.

[0032] Further, the communication between the gNB 106 and the UE 102 may be facilitated through messages transmitted at the data link layer (L2) and network layer (L3) of the communication network 104. In particular, the gNB 106 may receive the AI/ML model output parameters and/or KPIs and device capability information through messages transmitted at the data link layer (L2) and network layer (L3) of the communication network 104. In an aspect of the present disclosure, L2 or L3 may ensure reliable and efficient transmission between the network elements. Thus, ensuring that the communication between the gNB 106 and the UE 102 may adhere to the established protocols and standards governing cellular networks, consequently facilitating seamless data exchange, and enabling the effective monitoring and optimization of the AI/ML functionalities within the environment 100.

[0033] In an example, the gNB 106 may include one or more AI/ML models, utilized within the communication network 104, to optimize various functionalities such as CSI. In the example, a performance of the one or more AI/ML models may ensure efficient operation and enhance the overall quality of service within the environment 100. Accordingly, the AI/ML model output parameters and/or KPIs may refer to specific metrics or criteria that are predetermined or established to assess the performance of the one or more AI/ML models. Further, the KPIs may correspond to measurable parameters that indicate the performance or effectiveness of the one or more AI/ML models within the communication network 104. Thus, the KPIs may include metrics such as accuracy, precision, recall, convergence rate, computational efficiency, or any other relevant performance measure.

[0034] In an embodiment, the gNB 106 may be configured to receive the CSI feedback. The CSI feedback may include at least one of the pre-defined AI/ML output parameters and/or KPIs or the device capability information. In an example, the CSI feedback may correspond to data that characterizes the quality of the communication channel between the transmitter (such as the gNB 106) and the receiver (such as the UE 102). In an aspect of the present disclosure, the CSI feedback may optimize communication performance, resource allocation, and overall network efficiency.

[0035] In an example, the CSI feedback may include the AI/ML output parameters and/or KPIs to serve as benchmarks for evaluating the performance of the one or more AI/ML models within the gNB 106. In another example, the CSI feedback may include the device capability information referring to data related to the capabilities and characteristics of the UE 102 involved in the communication process. In the example, the device capability information may include information about the UE’s 102 hardware specifications, supported features, and communication protocols in the context of the CSI feedback.

[0036] In an embodiment, the gNB 106 may be configured to receive the AI/ML model output parameters and/or KPIs and the device capability information for the selected functionality or the use-case. In an example, the selected functionality or the use-case may include the CSI feedback, beam management, and positioning-related use-cases.

[0037] In an example, the selected functionality or the use-case including the CSI feedback may be adapted to optimize communication quality in the communication network 104. Consequently, incorporating the AI/ML model output parameters and KPIs and the device capability- information into the CSI feedback, allows network elements such as the gNB 106, to adapt transmission parameters for better performance. For instance, the one or more AI/ML models may predict channel behavior accurately, leading to better resource allocation and interference mitigation strategies.

[0038] In an example, the selected functionality or the use-case including positioning-related use-cases may correspond to accurate positioning of the UE 102 for various location-based services, including navigation, asset tracking, and emergency response. Thus, incorporating the AI/ML model output parameters and/or KPIs and the device capability information for the positioning-related use-cases may enhance positioning accuracy by learning from diverse data sources, such as signal strength measurements, time-of-arrival information, and environmental factors. In an aspect of the present disclosure, the gNB 106 may be robust and enable detection of reliable location estimation, even in challenging environments like remote locations or indoor areas.

[0039] In an embodiment, the gNB 106 may be configured to compare the AI/ML model output parameters and/or KPIs with corresponding predefined benchmarked values. In an example, the predefined benchmarked values may refer to target values or predefined thresholds against which the performance of the one or more AI/ML models may be assessed. Thus, the predefined benchmarked values may serve as reference points to estimate/evaluate the success or failure of the one or more AI/ML models of the gNB 106 in meeting the desired objectives for the functionalities within the communication network 104.

[0040] In an embodiment, the gNB 106 may configure the UE 102 upon determining that one or more of the AI/ML model output parameters and/or KPIs are below the corresponding predefined benchmarked value, thereby signifying that the gNB 106 may have evaluated the performance of the one or more AI/ML models based on the AI/ML model output parameters and/or KPIs received from the UE 102. Accordingly, in an example, if the actual performance, as indicated by the AI/ML model output parameters and/or KPIs may be less than the predefined benchmark values (pre-stored in the gNB 106), thus signifying that the one or more AI/ML models may not be meeting the desired objectives or required performance standards or having the sub-optimal performance. [0041] Consequently, upon determining that the one or more of the AI/ML model output parameters and/or KPIs are below the corresponding predefined benchmarked value, the gNB 106 may configure the UE 102 to perform non-AI/ML-based operations for the selected functionality or the use-case. In an example, the gNB 106 may instruct the UE 102 to switch from the AI/ML-based operations to the non-AI/ML-based operations for the selected functionality or the use-case. In the current example, the non-AI/ML-based operations may refer to conventional techniques that may not rely on artificial intelligence or machine learning techniques. For instance, the non-AI/ML-based operations may include predefined rules, heuristics, or algorithms. Further, once the gNB 106 configures the UE 102 to perform either the non-AI/ML-based operations or continue the AI/ML-based measurements and/or predictions, the UE 102 may be configured to communicate the UE’s 102 current operational mode (non-AI/ML-based operations or AI/ML-based predictions) back to the gNB 106. In an example, the UE 102 may communicate the current operational mode to the gNB 106 based on a 1-bit FLAG embedded within at least one of a data collection report, a measurement report message, or an AI/ML output delivery. In the example, the 1-bit FLAG may correspond to a binary⁷ indicator that may include two states i.e., 0 or 1. For instance, the 1-bit FLAG may represent whether the UE 102 is operating in the AI/ML-based measurements and/or predictions (1) or the non-AI/ML-based operation (0). The 1-bit FLAG enables the UE 102 to prevent the RACE condition by communicating a mode of prediction to the gNB 106.

[0042] Consequently, upon determining that one or more of the AI/ML model output parameters and/or KPIs are below the corresponding predefined benchmarked values, the gNB 106 may configure the UE 102 to disable or continue the AI/ML-based measurements and/or predictions for the selected functionality or the use-case based on the received device capability information.

[0043] In an example, the gNB 106 may be configured to determine factors such as a radio resource overhead at the gNB 106, an air interface overhead, resource consumption at the UE 102 and the gNB 106, and AI/ML-based capabilities of the UE 102. In the example, the radio resource overhead at the gNB 106 may correspond to the amount of radio resources such as bandwidth, frequency channels, etc., consumed by the gNB 106 to support the UE 102. Further, in the example, the air interface overhead may correspond to additional signaling or control messages required for communication over the air interface between the gNB 106 and the UE 102. Further, the resource consumption at the UE 102 and the gNB 106 may correspond to evaluating the resource utilization levels at both the UE 102 and the gNB 106, including processing power, memory usage, and energy consumption. Furthermore, the AI/ML-based capabilities of the UE 102 may correspond to the UE’s 102 ability’ to support the AI/ML-based functionalities, such as processing power, memory' capacity⁷, and compatibility with the one or more AI/ML models.

[0044] In an example, based on analyzing the factors, the gNB 106 may be configured to derive a priority value for the AI/ML-based measurements and/or predictions associated with the UE 102. In an example, the priority⁷ value may refer to the necessity' of continuing the AI/ML-based measurements and/or predictions associated with the selected functionality or use-case.

[0045] Consequently, in the example, the gNB 106 may be configured to continue the AI/ML- based measurements and/or predictions upon determining the derived priority⁷ value is more than a predetermined threshold value, associated with the selected functionality' or the use-case. Thus, indicating that continuing the AI/ML-based measurements and/or predictions may be beneficial or necessary for optimal network operations (the communication network 104).

[0046] Furthermore, in the example, the gNB 106 may be configured to receive an indication message from the UE 102, post receiving instructions to continue the AI/ML-based measurements and/or predictions. In the example, the indication message may refer to denote the UE’s 102 readiness to initiate the re-configuration process, i.e., enabling the AI/ML-based measurements and/or predictions for the selected functionality or the use-case. Furthermore, the reconfiguration process may also include disabling the non-AI/ML-based operations. In an aspect of the present disclosure, the indication message may serve as a signal that the UE 102 may be prepared to transition from the non-AI/ML-based operations to the AI/ML-based measurements and/or predictions for the selected functionality or the use-case. Consequently, in response to receiving the indication message from the UE 102, the gNB 106 may be configured to initiate the re-configuration process for enabling the AI/ML-based measurements and/or predictions for the selected functionality or use-case. In an aspect of the present disclosure, thus, the communication network 104 may ensure that the UE 102 operates optimally and efficiently, utilizing the benefits of the AI/ML techniques.

[0047] In an alternative example, the gNB 106 may configure the UE 102 to disable the AI/ML- based measurements and/or predictions for the selected functionality or the use-case, as the derived priority value may be less than the predetermined threshold value. This signifies that the benefits of the AI/ML-based measurements and/or predictions may be outweighed by associated overhead or resource constraints.

[0048] In the example, while the UE 102 may be operating with the AI/ML-based measurements and/or predictions being disabled, the gNB 106 may be configured to monitor pre-determined non- AI/ML-based KPIs/parameters associated with the non-AI/ML-based operations. The pre-determined non-AI/ML-based KPIs/parameters may be pre-stored in the gNB 106 to evaluate the perfonnance and efficiency of the network when operating without the one or more AI/ML models. In the present example, the gNB 106 may be configured to determine that the performance of the UE 102 does not meet the desired standards or objectives without the AI/ML-based measurements and/or predictions upon determining that predetermined non-AI/ML-based KPIs/parameters may be less than a corresponding non- AI/ML based predefined threshold value.

[0049] Simultaneously, the gNB 106 may also disable the non-AI/ML-based operations for the selected functionality or the use-case. In an aspect of the present disclosure, the disabling of the non-AI/ML-based operations may ensure that the UE 102 operates exclusively with the AI/ML- based measurements and/or predictions, thus, leveraging the benefits of machine learning and Al algorithms to optimize performance.

[0050] The communication network 104 includes one or more wireless networks. For example, the communication network 104 may include a cellular network (e.g., a 5G network, a Long- Term Evolution (LTE) network, a Third Generation (3G) network, a Code Division Multiple Access (CDMA) network, etc.), a Public Land Mobile Network (PLMN), a Local Area Network (LAN), a Wide Area Network (WAN), a Metropolitan Area Network (MAN), a private network, an ad hoc network, an intranet, the Internet, a Fiber optic-based network, or the like, and/or a combination of these or other types of networks.

[0051] The number and arrangement of devices and networks shown in FIG. 1 are provided as an example. In practice, there may be additional devices and/or networks, fewer devices and/or networks, different devices and/or networks, or differently arranged devices and/or networks than those shown in FIG. 1. Furthermore, two or more devices shown in FIG. 1 may be implemented within a single device, or a single device shown in FIG. 1 may be implemented as multiple, distributed devices. Additionally, or alternatively, a set of devices (e.g., one or more devices) of the environment 100 may perform one or more functions described as being performed by another set of devices of the environment 100.

[0052] FIG. 2 illustrates a functional framework 200 for an AI/ML model for a New Radio (NR) air interface, according to various embodiments of the present disclosure. The functional framework 200 includes a data collection component 202, a model training component 204, a management component 206, an inference component 208, and a model storage component 210.

[0053] The data collection component 202 is configured to collect raw data for the other component of the functional framework 200. The data collection component 202 may connected to one or more external devices to collect the raw data. The raw data may be processed by one or more other components of the functional framew ork 200 to obtain a desired result. For instance, the data collection component 202 may be configured to training data required for the operation of the model training component 204. The data collection component 202 may also be configured to collect monitoring data required for the operation of the management component 206. In other embodiments, the data collection component 202 maycollect inference data required for the operation of the inference component 208.

[0054] The model training component 204 may be configured to train the AI/ML model to predict a specific functionality or use-case. For instance, in the case of the NR air interface, the model training component 204 may be responsible for training the AI/ML model to effectively perform and analyze channel state information. In some embodiments, the model training component 204 may be configured to train the AI/ML model to perform CSI feedback enhancement, beam management, and positioning accuracy enhancement, as discussed herein.

[0055] The model training component 204 may be configured to implement different types of training modes. For instance, the model training component 204 may perform joint training of a two-side model, i.e., one AI/ML model being implemented at a UE 102 and another AI/ML model being implemented at the gNB 106. The joint-training may include training the two- sided model at a single entity, e.g.. at the UE 102 or the gNB 106. In another embodiment, the model training component 204 may perform training of the two-sided model at the UE 102 and the Gnb, respectively. In another embodiment, the model training component 204 may implement a separate training at the UE 102 and the gNB 106. In separate training, the UE- assisted CSI generation part and the gNB-assisted CSI reconstruction part are trained by the UE 102 and the gNB 106, respectively. In one embodiment, the UE 102 may correspond to the UE 102 (as shown in FIG. 1) and the gNB 106 may correspond to the gNB 106 (as shown in FIG. 1). In one example, the model training component 204 may receive performance feedback or a retraining request from the management component 206 to enhance the training of the AI/ML model.

[0056] The management component 206 corresponds to a function that oversees operations (e.g., selection/(de)activation/switching/fallback) and monitoring (e.g., performance) of AI/ML models or AI/ML functionalities. The management component 206 may also be configured to receive inference output from the inference component 208. The management component 206 may be configured to provide a management instruction to the inference component 208 based on the received inference output or the monitoring data from the data collection component 202. The management instruction may suggest at least one of one or more modifications and updates to the inference component 208 to ensure proper inference operation. In one embodiment, the management instruction may include information indicating at least one of selection, (de)activation, switching of the AI/ML models or the AI/ML-based functionalities, fallback to non-AI/ML operation (i.e., not relying on inference process), etc. [0057] The inference component 208 may be configured to receive inference data from the data collection component 202 and the management instruction from the management component 206. The inference component 208 may be configured to generate the required inference information. In an embodiment, such inference information may include the CSI feedback enhancement information, beam management information, and positioning accuracy enhancement information.

[0058] In an embodiment, the AI/ML models trained by the model training component 204 may be stored at the model storage component 210. The model storage component 210 may be configured to receive requests such as model transfer or delivery requests, from the management component 206. The model storage component 210 may transfer or deliver the required the AI/ML model(s) based on the received request from the management component 206. The model storage component 210 may transfer or deliver the required AI/ML model(s) to the inference component 208 to perform a desired inference operation. In particular, the inference component 208 may utilize the management instruction, the received AI/ML model(s), and the inference data to perform the desired inference operation.

[0059] In one embodiment, the various components of the functional framework 200 may be implemented as a set of instructions stored in a memory. In other embodiments, the various components of the function framework 200 may be implemented as sets of units, modules, hardware components, or a combination thereof.

[0060] FIG. 3 illustrates a sequence of operations between the UE 102 and the network entity (gNB) 106. according to various embodiments of the present disclosure. At operation 302. the gNB 106 may configure the UE 102 to perform measurement and reporting. In one embodiment, the measurement may correspond to a Radio Resource Measurement (RRM). In particular, at operation 302, the gNB 106 may transmit a management instruction to configure the UE 102 to perform measurement and reporting of the RRM. The RRM may correspond to, but is not limited to, CSI feedback enhancement, beam management, and positioning accuracy enhancements. The gNB 106 may configure the UE 102 to perform said measurement and reporting in accordance with exemplary use-cases. For instance, the gNB 106 may configure the UE 102 to perform measurement and reporting after every predefined interval of time. Alternatively, the gNB 106 may configure the UE 102 to perform measurement and reporting after an occurrence of an event such as, but is not limited to, receiving a management instruction from the gNB 106.

[0061] At operation 304, the UE 102 may transmit at least one of a performance and assistance information based on the performed measurement. The at least one of the performance and assistance information may include, but is not limited to, Channel Quality⁷ Indicators (CQIs), Reference Signal Received Power (RSRP), Reference Signal Received Quality (RSRQ), Buffer Status Reports (BSR), positioning information, interference measurements, and so forth. The CQI may indicate a quality of the downlink channel. The CQI may assist the gNB 106 to update transmission parameters, such as modulation and coding schemes, to optimize a communication link between the UE 102 and the gNB 106. The RSRP and RSRQ may indicate a strength and a quality of the signals received from the gNB 106. The RSRP and RSRQ may assist the gNB 106 in making appropriate handover decisions and power control operations. The BSR may indicate an amount of data in a queue for the transmission. The BSR may assist the gNB 106 to determine resource allocation and scheduling operations. The positioning information may indicate a current position of the UE 102 and may assist the gNB 106 to provide location-based services to the UE 102. The interference measurements may indicate interference experience on different channels and frequencies established between the UE 102 and the gNB 106. The UE 102 may also indicate whether such measurement has been performed using the AI/ML- based prediction or the non-AI/ML-based operations.

[0062] At operation 306, a management function at the gNB 106 analyses the received at least one of performance and assistance information from the UE 102 and generates management instructions. The management instructions may correspond to the configuration of the UE 102 indicating whether the UE 102 needs to perform the AI/ML-based prediction or the non-AI/ML- based operations.

[0063] At operation 308, the gNB 106 transmits the management instructions to the UE 102.

Thus, the gNB 106 may configure the UE 102 to perform measurement and reporting according to the determined AI/ML-based prediction or the non-AI/ML-based operations using the management instructions. In one embodiment, the various management instructions may be transmitted by the gNB 106 using Radio Resource Control (RRC) or Media Access Control (MAC) signaling.

[0064] FIG. 4 illustrates another sequence of operations between the UE 102 and the network entity (gNB) 106, according to various embodiments of the present disclosure. Similar to FIG. 3, at operation 402, the gNB 106 may configure the UE 102 to perform measurement and reporting. However, in FIG. 4, at operation 404, a management function at the UE 102 analyses the measured at least one of performance and assistance information and generates a management request. The management request may indicate that the UE 1 2 requires switching from either the AI/ML-based prediction to the non-AI/ML-based operation or vice-versa.

[0065] At operation 406, the UE 102 may transmit the generated management request to the gNB 106. Upon receiving the management request, the gNB 106 determines whether to perform the requested switching. At operation 408, the gNB 106 transmits the management instructions to the UE 102 in response to the received management request. The management instructions may either allow the UE 102 to switch to a desired mode of operation i.e., the requested switching, or may reject the requested switching. In one embodiment, the various management instructions may be transmitted by the gNB 106 using a Radio Resource Control (RRC) or a Media Access Control (MAC) signaling.

[0066] FIG. 5 illustrates a block diagram of the base station (gNB 106), according to various embodiments of the present disclosure. The base station 106 includes a transceiver 502, a memory 504, one or more processors 506 (hereinafter referred to as the processor 506), and a plurality of modules 508. The transceiver 502, the memory 504, the processor 506, and the plurality of modules 508 may be communicably coupled to each other.

[0067] The transceiver 502 may be configured for transmitting and receiving wireless signals from and at the gNB 106. In one embodiment, the transceiver 502 may facilitate communication between the UE 102 and the gNB 106 via the communication network 104. In an example embodiment, the transceiver 502 may be configured for the transmission of data from the gNB 106 to the UE 102, as well as the reception of data from the UE 102 back to the gNB 106. The bidirectional communication enables various services such as voice calls, video streaming, and internet browsing. In an embodiment, the bidirectional communication enables one or more measurement functionalities of the NR including RRM. The RRM may correspond to, but is not limited to, CSI feedback enhancement, beam management, and positioning accuracy enhancements.

[0068] The processor 506 may include one or more processing units or other processing devices that control the overall operation of the gNB 106. As an example, the processor 506 may be a single processing unit or a number of units, one or more of which could include multiple computing units. The processor 506 may be implemented as one or more microprocessors, microcomputers, microcontrollers, digital signal processors, central processing units, state machines, logic circuitries, or any devices that manipulate signals based on operational instructions. Among other capabilities, the processor 506 is configured to fetch and execute computer-readable instructions and data stored in the memory 504. The processor 506 may include one or a plurality of processors. At this time, one or a plurality of processors may be a general-purpose processor, such as a central processing unit (CPU), an application processor (AP), or the like.

[0069] The memory 504 may include any non-transitory computer-readable medium known in the art including, for example, one or more of volatile memory. such as Static Random Access Memory (SRAM) and Dynamic Random Access Memory (DRAM), and non-volatile memory, such as Read-Only Memory (ROM), erasable programmable ROM, flash memones, hard disks, optical disks, and magnetic tapes. The non-transitory computer-readable medium may include a set of instructions that when executed cause the processor 506 to perform the method as discussed herein.

[0070] In one embodiment, the processor 506 may be configured to receive one or more UE- based AI/ML model output parameters and/or KPIs (interchangeably referred to as ‘ AI/ML model KPIs”) and device capability information from the UE 102. In one embodiment, the processor 506 may transmit a request to receive the one or more UE-based AI/ML model output parameters and/or KPIs and device capability information from the UE 102. In another embodiment, the processor 506 may periodically receive the one or more UE-based AI/ML model output parameters and/or KPIs and device capability⁷ information from the UE 102 over a predefined interval of time. In one embodiment, the predefined interval of time may be based on implementation of the RRM procedure. In yet another embodiment, the processor 506 may define the predefined interval of time based on real-time data as received in response to the implemented RRM procedure. Examples of the one or more UE-based AI/ML model output parameters and/or KPIs may include, but are not limited to, an accuracy, a precision, and recall, a Mean Absolute Error (MAE), a Root Mean Squared Error (RMSE), a Mean Average Precision (mAP), a confusion matrix, model interpretability⁷, and the like. The UE-based AI/ML model output parameters and/or KPIs may provide a comprehensive overview of the performance of the AI/ML model implemented at the UE 102. The performance of the AI/ML model may encompass aspects of accuracy, reliability, efficiency, fairness, and interpretability. The device capability information associated with the UE 102 may include, but is not limited to, Operation System (OS), processor’s configuration, memory space, supported network bands, maximum data transfer rate, Quality of Service (QoS) parameters, network-related parameters, and so forth.

[0071] The one or more UE-based AI/ML model output parameters and/or KPIs and the device capability information may be received for one or more selected functionality⁷ or use-case. Examples of the one or more selected functionality⁷ or use-case may include, but are not limited to, CSI feedback, beam management, and positioning-related use-cases. The CSI feedback is responsible for minimizing overhead associated with exchanging channel state information between the UE 102 and the gNB 106. The CSI feedback is a mechanism used in wireless communication systems, particularly in cellular networks, to provide information related to a current state of a communication channel established between the gNB 106 (i.e., the transmitter) and the UE 102 (i.e., a receiver). In wireless communication, the channel state refers to the conditions of the wireless transmission medium that may be affected by factors such as, but not limited to, signal attenuation, multipath propagation, interference, and fading. Therefore, the channel state varies over time and space, making it essential for the transmitter to adapt transmission parameters to optimize signal quality and reliability⁷. The CSI feedback enables the gNB 106 to optimize the transmission parameters for the UE 102. The CSI feedback may include information such as, but not limited to, channel gains, Signal-to-Noise Ratio (SNR), Signal-to-Interference-plus-Noise Ratio (SINR), and other parameters characterizing the quality and reliability of the received signal. The beam management may correspond to an operation of effectively managing and optimizing utilization of direction beams in wireless communication systems. The beam management may encompass information such as, but not limited to, channel conditions, mobility, network topology required for beam prediction, directional information of transmitted and received beams required for beam steering, signal quality, interference levels, and resource availability required for beam selection, and coverage area, data rate, and interference mitigation required for beam-forming mode selection, and so forth. At least one of the UE 102 and the gNB 106 may implement one or more AI/ML-based models to perform the above-discussed functionalities of the wireless communication systems. The one or more UE-based AI/ML model output parameters and/or KPIs may indicate aspects of accuracy, reliability, efficiency, fairness, and interpretability, associated with the abovediscussed functionalities.

[0072] The processor 506 may also be configured to compare the one or more UE-based AI/ML model output parameters and/or KPIs with corresponding predefined benchmarked values. The predefined benchmarked values may define a minimum parameter value required for a KPI to obtain a desired result. In one non-limiting example, the accuracy KPI may have a predefined benchmark value of 70%. However, the predefined benchmarked value may be defined in any suitable manner including a definite value, a percentage, a range of values, and the like. Similarly, the predefined benchmarked value may be defined for one or more of the UE-based AI/ML model output parameters and/or KPIs. The processor 506 may compare the one or more UE-based AI/ML model output parameters and/or KPIs with the corresponding predefined benchmarked value. In one embodiment, the predefined benchmarked value may be defined by a user (for example, a network operator) based on the requirements of the network. In another embodiment, the predefined benchmarked value may be defined by the processor 506 based on a desired operation of the gNB 106. [0073] Moreover, the processor 506 may configure the UE 102 to perform non-AI/ML-based operations for the one or more of selected functionality or use-case, upon determining that at least one of the one or more UE-based AI/ML model output parameters and/or KPIs is below the corresponding predefined threshold value. For instance, if the processor 506 determines that the accuracy KPI is below 70%, the processor 506 may configure the UE 102 to fallback to the non-AI/ML-based operations for the one or more of the selected functionality or use-cases. In one embodiment, the processor 506 may configure the UE 102 to perform one of disabling or continuing AI/ML-based measurements and/or predictions for the one or more of selected functionality or use-case based on one or more predefined criteria corresponding to the received device capability information. The one or more predefined criteria may include a correlation between the received device capability information and one or more air interface overhead, signaling overhead, UE power consumption, UE subscription type, UE and gNB processing overhead, and additional radio resource overhead caused by the AI/ML based measurements and/or predictions. The processor 506 may configure the UE 102 to perform one of disabling or continuing AI/ML-based measurements and/or predictions, upon determining that at least one of the one or more UE-based AI/ML model output parameters and/or KPIs is below the corresponding predefined threshold value. In one embodiment, the processor 506 may implement the management function (as discussed in reference to FIGS. 2-4) to perform the configuration of the UE 102. In particular, the processor 506 may transmit a management instruction to configure the UE 102 based on the one or more UE-based AI/ML model output parameters and/or KPIs and the device capability information. The management instruction may define a set of operations for the UE 102.

[0074] In one embodiment, to configure the UE 102 to perform one of disable or continue AI/ML-based prediction, the processor 506 may be configured to determine one or more of a radio resource overhead at the gNB 106, an air interface overhead, resource consumption at the UE 102 and the gNB 106. and AI/ML-based capabilities of the UE 102. The radio resource overhead at the gNB 106 may correspond to radio resources required by the gNB 106 to perform various signaling, control, and management functions. Similarly, the air interface overhead may correspond to additional signaling and control information transmitted over a wireless communication channel between the UE 102 and the gNB 106 beyond payload data. The resource consumption at the UE 102 and the gNB 106 may indicate real-time resource utilization at the UE 102 and the gNB 106, respectively. Further, the AI/ML-based capabilities of the UE 102 may correspond to a device capability of the UE 102 to support operation of the one or more AI/ML models implemented at the UE 102.

[0075] The processor 506 may also be configured to analyze the determined radio resource overhead at the gNB 106, air interface overhead, resource consumption at the UE 102 and the gNB 106, and AI/ML-based capabilities ofthe UE 102. The processor 506 may derive a priority of the AI/ML-based measurements and/or predictions for the UE 102 based on said analysis. The priority may indicate one or more requirements or support for the AI/ML-based prediction for the UE 102. For instance, in case the UE 1 2 has low AI/ML-based capabilities, the priority for AI/ML-based measurements and/or predictions for the UE 102 may also be low. Similarly, if the UE 102 has high AI/ML-based capabilities, the priority for AI/ML-based measurements and/or predictions for the UE 102 may also be high. In a similar manner, the determined radio resource overhead at the gNB 106, the air interface overhead, and the resource consumption at the UE 102 and the gNB 106 may impact the priority’ for the AI/ML-based prediction for the UE 102. The impact of the detennined radio resource overhead at the gNB 106, the air interface overhead, and the resource consumption at the UE 102 and the gNB 106 may be either directly or indirectly proportional to the priority of the AI/ML-based prediction. The processor 506 may also analyze the determined priority of the AI/ML-based prediction with respect to a predetermined threshold value. One non-limiting example of such a predetermined threshold value may be 85% or 0.8.

[0076] Further, in response to determining that the derived priority is above a predetermined threshold value, the processor 506 may be configured to configure the UE 102 to continue the AI/ML-based measurements and/or predictions for the one or more of the selected functionality or use-case. Also, in response to determining that the derived priority is below the predetermined threshold value, the processor 506 may be configured to configure the UE 102 to disable the AI/ML-based measurements and/or predictions for the at least one selected functionality or use-case. In particular, based on the comparison of the derived priority of the AI/ML-based prediction with respect to the corresponding predetermined threshold value, the processor 506 may determine whether to disable or continue the AI/ML-based prediction at the UE 102.

[0077] In response to configuring the UE 102 to continue the AI/ML-based measurements and/or predictions for the at least one selected functionality or use-case, the processor 506 may be configured to receive an indication message from the UE 102. The processor 506 may receive the indication message based on the one or more pre-determined KPIs or parameters satisfying pre-determined threshold values. The indication message may initiate re-configuring the AI/ML-based measurements and/or predictions for the at least one selected functionality or use-case and disable the non- AI/ML-based operations. In particular, the UE 102 may monitor and analyze radio resource overhead at the gNB 1 6, the air interface overhead, the resource consumption at the UE 102, and the gNB 106 corresponding to AI/ML-based prediction. The UE 102 may determine whether to re-initiate the AI/ML-based prediction and disable the non- AI/ML operation based on said analysis. Thereafter, the UE 102 may indicate the determined output to the gNB 106 via the indication message. The gNB 106 may analyze the received indication message and decide whether to re-configure the AI/ML-based measurements and/or predictions and disable the non- AI/ML-based operations for the UE 102.

[0078] In response to configuring the UE 102 to disable the AI/ML-based measurements and/or predictions for the at least one selected functionality or use-case, the processor 506 may be configured to monitor one or more pre-determined non-AI/ML-based KPIs/parameters. Moreover, the processor 506 may be configured to enable the AI/ML-based measurements and/or predictions for the one or more of the selected functionality or use-case upon determining that at least one of the one or more pre-determined non- AI/ML-based KPIs/parameters is below a corresponding non-AI/ML based predefined threshold value. The processor 506 may also be configured to disable the non-AI/ML-based operations upon determining that at least one of the one or more pre-determined non-AI/ML-based KPIs/parameters is below a corresponding non- AI/ML-based predefined threshold value. In particular, similar to UE-based AI/ML model output parameters and/or KPIs analysis, the processor 506 may also perform analysis of the non-AI/ML-based KPIs/parameters and corresponding predefined threshold values. Based on said analysis, the processor 506 may determine whether to continue non-AI/ML-based operation or switch back to the AI/ML-based measurements and/or predictions.

[0079] In some embodiments, the processor 506 may be configured to configure the UE 102 to indicate one of the AI/ML-based measurements and/or predictions or the non-AI/ML- based operations for the at least one selected functionality or use-case using a 1 -bit FLAG in at least one of a data collection report, a measurement report message, or an AI/ML output delivery message. The indication may assist the processor 506 to correctly identify if the measurements are being performed using the AI/ML-based prediction or the non-AI/ML-based operations. This may assist the processor 506 to effectively compare the corresponding KPIs and thresholds to determine appropriate measurement decisions.

[0080] In one embodiment, the processor 506 may be configured to receive a channel state information (CSI) feedback comprising at least one of the one or more UE-based AI/ML model output parameters and/or KPIs or the device capability information.

[0081] In some embodiments, the one or more modules 508 may include a set of instructions that may be executed to cause the gNB 106 to perform any one or more of the methods /processes disclosed herein. The one or more modules 508 may be configured to perfonn the steps of the present disclosure using the data stored in the memory 504 or data received from the UE 102. In some embodiments, the one or more modules 508 may be stored within the memory 504. In an embodiment, the one or more modules 508 may be a hardware unit that may be outside the memory 504.

[0082] In an example embodiment, the one or more modules 508 may include an AI/ML module 510. The AI/ML module 510 may be configured to implement one or more AI/ML models. The one or more AI/ML models may be configured to generate various management functions required for interworking of the AI/ML-based prediction and non-AI/ML-based operations at the UE 102. [0083] In one embodiment, the one or more modules 508 may be implemented through an Al model. The Al model may be defined as a function associated with Al and may be performed through the non-volatile memory', the volatile memory', and the processor 506.

[0084] The processor 506 may include one or a plurality of processors. At this time, one or a plurality of processors may be a general purpose processor, such as a Central Processing Unit (CPU), an Application Processor (AP), or the like, a graphics-only processing unit such as a Graphics Processing Unit (GPU), a Visual Processing Unit (VPU), and/or an Al-dedicated processor such as a Neural Processing Unit (NPU).

[0085] The one or a plurality of processors control the processing of the input data in accordance with a predefined operating rule or the Al model stored in the non-volatile memory and/or the volatile memory'. In an embodiment, the predefined operating rule or the Al model is provided through training or learning.

[0086] Here, being provided through learning means that, by applying a learning technique to a plurality of learning data, the predefined operating rule or the Al model of a desired characteristic is made. The learning may be performed in a device itself in which Al according to an embodiment is performed, and/or may be implemented through a separate server/system.

[0087] The Al model may consist of a plurality of neural network layers. Each layer has a plurality of weight values and performs a layer operation through the calculation of a previous layer and an operation of a plurality of weights. Examples of neural networks include, but are not limited to, Convolutional Neural Network (CNN), Deep Neural Network (DNN), Recurrent Neural Network (RNN), Restricted Boltzmann Machine (RBM), Deep Belief Network (DBN). Bidirectional Recurrent Deep Neural Network (BRDNN), Generative Adversarial Networks (GAN), and deep Q-networks.

[0088] The learning technique is a method for training a predetermined target device (for example, the gNB and/or the UE) using a plurality of learning data to cause, allow, or control the target device to make a determination or prediction. Examples of learning techniques include, but are not limited to, supervised learning, unsupervised learning, semi-supervised learning, or reinforcement learning.

[0089] The Al model may be obtained by training. Here, "obtained by training" means that the predefined operation rule or the Al model configured to perform a desired feature (or purpose) is obtained by training a basic Al model with multiple pieces of training data by a training technique.

[0090] FIGS. 6a-6c illustrate a flow chart of an example method 600, in accordance with various embodiments of the present disclosure. The method 600 may be performed by the gNB 106, also referred to as the apparatus 106.

[0091] At operation 602, the apparatus 106 may receive the one or more UE-based AI/ML model output parameters and/or KPIs and device capability infonnation from the UE 102. The one or more UE-based AI/ML model output parameters and/or KPIs and device capability information may correspond to at least one selected functionality or use-case. Examples of the selected functionality or use-case include, but are not limited to, CSI feedback, beam management, and positioning-related use-cases.

[0092] At operation 604, the apparatus 106 may compare the one or more UE-based AI/ML model output parameters and/or KPIs with corresponding predefined benchmarked values. In response to determining that one or more UE-based AI/ML model output parameters and/or KPIs are below the corresponding predefined benchmarked values, the apparatus 106 may perform the operations 606 and 608.

[0093] At operation 606. the apparatus 106 configures the UE 102 to perform fallback to non- AI/ML-based operations for the at least one selected functionality or use-case. Thus, the apparatus 106 may reduce the overhead at the network resources. At operation 608, the apparatus 106 may determine a radio resource overhead at the gNB 106, an air interface overhead, resource consumption at the UE 102 and the gNB 106, and AI/ML-based capabilities of the UE 102. Next, at operation 610, the apparatus 106 may analyze the determined radio resource overhead at the gNB 106, air interface overhead, resource consumption at the UE 102 and the gNB 106, and AI/ML-based capabilities of the UE 102 and derive a priority of the AI/ML-based measurements and/or predictions for the UE 102. Based on said analysis and the priority of the AI/ML-based measurements and/or predictions for the UE 102, the apparatus 106 may either perform operation 612 or operation 614. Specifically, in response to determining that the derived priority is above a predetermined threshold value, the apparatus 106 may perform the operation 612. However, in response to determining that the derived priority is below the predetermined threshold value, the apparatus 106 may perform the operation 614.

[0094] At operation 612. the apparatus 106 may configure the UE 102 to continue the AI/ML- based measurements and/or predictions for the at least one selected functionality or use-case. Thereafter, at operation 616, the apparatus 106 may receive an indication message from the UE 102. The indication message may be based on one or more pre-determined KPIs or parameters satisfying pre-determined threshold values. The indication message may initiate re-configuring the AI/ML-based measurements and/or predictions for the at least one selected functionality or use-case and disable the non- AI/ML-based operations. Thus, the apparatus 106 may effectively enable the AI/ML-based measurements and/or predictions for the UE 102 that can support both non- AI/ML-based operations and the AI/ML-based measurements and/or predictions.

[0095] At operation 614, the apparatus 106 may configure the UE 102 to disable the AI/ML- based measurements and/or predictions for the at least one selected functionality or use-case. Thereafter, at operation 618, the apparatus 106 may monitor one or more pre-determined non- AI/ML-based KPIs/parameters. Further, at operation 620, the apparatus 106 enables the AI/ML-based measurements and/or predictions for the at least one selected functionality or usecase. The apparatus 106 may enable the AI/ML-based measurements and/or predictions upon determining that at least one of the one or more pre-determined non-AI/ML-based KPIs/parameters is below a corresponding non-AI/ML based predefined threshold value. Moreover, the apparatus 106 may also disable the non-AI/ML-based operations upon determining that at least one of the one or more pre-determined non-AI/ML-based KPIs/parameters is below a corresponding non-AI/ML-based predefined threshold value. [0096] At operation 608, the apparatus 106 may determine a radio resource overhead at the gNB 106, an air interface overhead, resource consumption at the UE 102 and the gNB 106, and AI/ML-based capabilities of the UE 102. Next, at operation 610, the apparatus 106 may analyze the determined radio resource overhead at the gNB 106, air interface overhead, resource consumption at the UE 102 and the gNB 106, and the AI/ML-based capabilities of the UE 102. The apparatus 106 may derive a priority of the AI/ML-based measurements and/or predictions for the UE 102.

[0097] Moreover, in response to determining that the derived priority is above a predetermined threshold value, the apparatus 106 configures the UE 102 to continue the AI/ML-based measurements and/or predictions for the at least one selected functionality or use-case, as shown in operation 612. After configuring the UE 102 to continue the AI/ML-based measurements and/or predictions, the apparatus 106 may receive, from the UE 102, an indication message. The indication message may be based on one or more pre-determined KPIs or parameters satisfying pre-determined threshold values. The indication message may be transmitted by the UE 102 to initiate re-configuring the AI/ML-based measurements and/or predictions for the at least one selected functionality or use-case and disable the non- AI/ML-based operations.

[0098] FIG. 7 illustrates a flow chart of an example method 700, in accordance with various embodiments of the present disclosure. The method 700 may be performed by the gNB 106, also referred to as the apparatus 106.

[0099] In operation 702, the method 700 includes receiving, by the gNB 106, one or more UE- based AI/ML model output parameters and/or KPIs and device capability information. The one or more UE-based AI/ML model output parameters and/or KPIs and device capability information are received from the UE 102 for at least one selected functionality or use-case.

[00100] In operation 704, the method 700 includes comparing, by the gNB 106, the one or more UE-based AI/ML model output parameters and/or KPIs with corresponding predefined benchmarked values. [00101] In operation 706, the method 700 includes configuring, by the gNB 106, the UE 102 to fallback to non-AI/ML-based operations for the at least one selected functionality or usecase. The method also includes configuring, by the gNB 106, the UE 102 to perform one of disabling or continuing AI/ML-based measurements and/or predictions for the at least one selected functionality or use-case. The one of disabling or continuing the AI/ML-based measurements and/or predictions is performed based on one or more predefined criteria corresponding to the received device capability information. The operation 706 is performed upon determining that at least one of the one or more AI/ML model KPIs is below the corresponding predefined benchmarked value.

[00102] Embodiments are exemplary in nature and the sequence of the method 700 may vary with respect to omission or change in sequence of one or more steps.

[00103] FIG. 8 illustrates an embodiment of a device/apparatus 800. The device 800 may correspond to any one of the UE 102 and the base station 106, as shown in FIG. 1. As shown in FIG. 8, the device 800 may include a processor 810, a memory 820, a storage component 830, an input component 840, an output component 850, a communication interface 860, and a bus 870.

[00104] The processor 810, as used herein, means any type of computational circuit that may comprise hardware elements and software elements. The processor 810 may be embodied as a multi-core processor, a single-core processor, or a combination of one or more multi-core processors and one or more single-core processors, a distributed processing system, or the like. The processor 810 may be a Central Processing Unit (CPU) a graphics processing unit (GPU), an accelerated processing unit (APU), an application-specific integrated circuit (ASIC), or another type of processing component.

[00105] The memory' 820 includes a random-access memory (RAM), a read-only memory (ROM), and/or another type of dynamic or static storage device (e.g., a flash memory, a magnetic memory, and/or an optical memory) that stores information and/or instructions for use by processor 810. The memory' 820 comprises machine-readable instructions which are executable by the processor 810. These machine-readable instructions when executed by the processor 810 cause the processor 810 to perform method steps of an exemplary embodiment described herein.

[00106] The storage component 830 stores information and/or software related to the operation and use of the device 800. For example, the storage component 830 may include a hard disk (e.g., a magnetic disk, an optical disk, a magneto-optic disk, and/or a solid-state disk), a compact disc (CD), a digital versatile disc (DVD), a floppy disk, a cartridge, a magnetic tape, and/or another type of non- transit ory computer-readable medium, along with a corresponding drive.

[00107] The input component 840 is configured to receive information, such as via user input. For example, the input component 840 may include, but not be limited to, a touch screen display, a keyboard, a keypad, a mouse, a button, a switch, and/or a microphone. Additionally, or alternatively, the input component 840 may include a sensor for sensing information (e.g., a global positioning system (GPS), an accelerometer, a gyroscope, and/or an actuator).

[00108] The output component 850 is configured to provide output information from the device Y00. For example, the output component 850 may be, but is not limited to, a display, a speaker, and/or one or more light-emitting diodes (LEDs).

[00109] The communication interface 860 is an interface that provides a communication connection to other devices. The connection by the communication interface 860 can be a wired connection, a wireless connection, or a combination of wired and wireless connections, and can be a direct connection or an indirect connection via a communication network 104 that exists between other devices. In other words, the standard of the communication interface 860 is not limited.

[00110] The bus 870 acts as an interconnect between the processor 810, the memory 820, the storage component 830, the input component 840, the output component 850, and the communication interface 860 of the device 800. [00111] The number and arrangement of components shown in FIG. 8 are provided as an example. In practice, the device 800 may include additional components, fewer components, different components, or differently arranged components than those shown in FIG. 8. Additionally, or alternatively, a set of components (e.g., one or more components) of device 800 may perform one or more functions described as being performed by another set of components of device 800.

[00112] According to one aspect, a method is provided. The method includes receiving, by a gNB, one or more UE-based artificial intelligence/machine learning (AI/ML) model output parameters and/or key performance indicators (KPIs). The method also includes receiving device capability information. The one or more UE-based AI/ML model output parameters and/or KPIs and device capability information is received from a user equipment (UE) for at least one selected functionality' or use-case. The method includes comparing, by the gNB, the one or more UE-based AI/ML model output parameters and/or KPIs with corresponding predefined benchmarked values. The method further includes upon determining that at least one of the one or more UE-based AI/ML model output parameters and/or KPIs is below' the corresponding predefined benchmarked value configuring, by the gNB, the UE. The method includes configuring the UE to perform non-AI/ML-based operations for the at least one selected functionality or use-case. Furthermore, the method includes configuring the UE to perform one of disabling or continuing AI/ML-based measurements and/or predictions for the at least one selected functionality' or use-case. The UE is configured to perform one of disabling or continuing AI/ML-based measurements and/or predictions based on one or more predefined criteria corresponding to the received device capability information.

[00113] The method described in para [0111], further includes determining, by the gNB, a radio resource overhead at the gNB, an air interface overhead, resource consumption at the UE and the gNB, and AI/ML-based capabilities of the UE. The method also includes analyzing, by the gNB, the determined radio resource overhead at the gNB, air interface overhead, resource consumption at the UE and the gNB, and AI/ML-based capabilities of the UE. The method includes deriving a priority of the AI/ML-based measurements and/or predictions for the UE. In response to determining that the derived priority is above a predetermined threshold value, the method includes configuring, by the gNB, the UE to continue the AI/ML-based measurements and/or predictions. The AI/ML-based measurements and/or predictions are for the at least one selected functionality or use-case. In response to determining that the derived priority is below the predetermined threshold value, the method includes configuring, by the gNB. the UE to disable the AI/ML based measurements and/or predictions. The AI/ML based measurements and/or prediction is for the at least one selected functionality' or use-case.

[00114] The method described in any one of paragraphs [0111] - [0112], further includes monitoring, by the gNB, one or more pre-determined non- AI/ML based KPIs/parameters. The method includes monitoring in response to configuring the UE to disable the AI/ML based measurements and/or predictions for the at least one selected functionality or use-case. The method also includes enabling, by the gNB, the AI/ML based measurements and/or predictions for the at least one selected functionality or use-case. The AI/ML based measurements and/or predictions are enabled upon determining that at least one of the one or more pre-determined non-AI/ML based KPIs/parameters is below a corresponding non-AI/ML based predefined threshold value. The method also includes disabling the non-AI/ML based operations. The non- AI/ML based operations are disabled upon determining that at least one of the one or more predetermined non-AI/ML based KPIs/parameters is below a corresponding non-AI/ML based predefined threshold value.

[00115] The method described in any one of paragraphs [0111] - [0113], the method also includes configuring, by the gNB, the UE to indicate one of the AI/ML-based measurements and/or predictions or the non-AI/ML- based operations. The AI/ML-based measurements and/or predictions or the non-AI/ML-based operations are for the at least one selected functionality or use-case. The indication is provided using a 1 -bit FLAG in a data collection report, a measurement report message, or an AI/ML output delivery message.

[00116] The method described in any one of paragraphs [0111] - [0114], includes receiving a channel state information (CS1) feedback comprising at least one of the one or more UE-based AI/ML model output parameters and/or KPIs or the device capability information. [00117] The method described in any one of paragraphs [0111] - [01 15], the at least one selected functionality or use-case comprises one or more of channel state information (CSI) feedback, beam management, and positioning-related use-cases.

[00118] The method described in any one of paragraphs [0111] - [0116], is performed using an L2/L3 message.

[00119] The method described in any one of paragraphs [0111] - [0117], the one or more predefined criteria includes a correlation between the received device capability information and one or more air interface overhead, signaling overhead, UE power consumption, UE subscnption type, UE and gNB processing overhead, and additional radio resource overhead caused by the AI/ML based measurements and/or predictions.

[00120] According to another aspect, an apparatus is disclosed. The apparatus is configured to receive one or more UE-based artificial intelligence/machine learning (AI/ML) model output parameters and/or key performance indicators (KPIs). The apparatus is also configured to receive device capability information. The one or more UE-based AI/ML model output parameters and/or KPIs and the device capability information are received from a user equipment (UE) for at least one selected functionality or use-case. The apparatus is also configured to compare the one or more UE-based AI/ML model output parameters and/or KPIs wi th corresponding predefined benchmarked values. Upon determining that at least one of the one or more UE-based AI/ML model output parameters and/or KPIs is below the corresponding predefined benchmarked value, the apparatus is configured to configure the UE. The apparatus is configured to configure the UE to perform non-AI/ML-based operations for the at least one selected functionality or use-case. The apparatus is also configured to configure the UE to perform to perform one of disabling or continuing AI/ML-based measurements and/or predictions for the at least one selected functionality or use-case based on one or more predefined criteria corresponding to the received device capability information.

[00121] The apparatus described in para [01 19], is further configured to determine a radio resource overhead at the gNB, an air interface overhead, resource consumption at the UE and the gNB. The apparatus is also configured to determine the AI/ML-based capabilities of the UE. The apparatus is further configured to analyze the determined radio resource overhead at the gNB, air interface overhead, and resource consumption at the UE and the gNB. The apparatus is also configured to analyze the AI/ML-based capabilities of the UE. Moreover, the apparatus is configured to derive a priority of the AI/ML-based measurements and/or predictions for the UE. In response to determining that the derived priority is above a predetermined threshold value, the apparatus is configured to configure the UE to continue the AI/ML-based measurements and/or predictions. The AI/ML-based measurements and/or predictions are for the at least one selected functionality or use-case. In response to determining that the derived priority is below the predetermined threshold value, the apparatus is configured to configure the UE to disable the AI/ML based measurements and/or predictions. The AI/ML based measurements and/or prediction is for the at least one selected functionality' or use-case.

[00122] The apparatus described in any one of paragraphs [0119] - [0120], is further configured to monitor one or more pre-determined non- AI/ML based KPIs/parameters. The monitoring is performed in response to configuring the UE to disable the AI/ML based measurements and/or predictions for the at least one selected functionality' or use-case. The apparatus is also configured to enable the AI/ML based measurements and/or predictions for the at least one selected functionality or use-case. The AI/ML based measurements and/or predictions is enabled upon determining that at least one of the one or more pre-determined non- AI/ML based KPIs/parameters is below a corresponding non- AI/ML based predefined threshold value. The apparatus is also configured to disable the non- AI/ML based operations. The non-AI/ML based operations are disabled upon determining that at least one of the one or more pre-determined non-AI/ML based KPIs/parameters is below a corresponding non-AI/ML based predefined threshold value.

[00123] The apparatus described in any one of paragraphs [0119] - [0121], is configured to configure the UE to indicate one of the AI/ML-based measurements and/or predictions or the non-AI/ML- based operations. The AI/ML-based measurements and/or predictions or the non- AI/ML- based operations is for the at least one selected functionality or use-case. The indication is provided using 1 -bit FLAG in a data collection report, a measurement report, or an AI/ML output delivery message.

[00124] The apparatus described in any one of paragraphs [0119] - [0122], is configured to receive a channel state information (CSI) feedback comprising at least one of the one or more UE-based AI/ML model output parameters and/or KPIs or the device capability information.

[00125] According to another aspect, a non-transitory computer-readable medium storing instructions is disclosed. The instructions includes one or more instructions that are executed by a network device comprising one or more processors. The instructions cause the one or more processors to receive one or more UE-based artificial intelligence/machine learning (AI/ML) model output parameters and/or key performance indicators (KPIs) and device capability information. The one or more UE-based AI/ML model output parameters and/or KPIs and the device capability information are received from a user equipment (UE) for at least one selected functionality or use-case. The instructions cause the one or more processors to compare the one or more UE-based AI/ML model output parameters and/or KPIs with corresponding predefined benchmarked values. Upon determining that at least one of the one or more UE-based AI/ML model output parameters and/or KPIs is below the corresponding predefined benchmarked value, the instructions cause the one or more processors to configure the UE. The UE is configured to perform non-AI/ML-based operations for the at least one selected functionality or use-case. The UE is also configured to perform one of disabling or continuing AI/ML-based measurements and/or predictions for the at least one functionality or use-case based on one or more predefined criteria corresponding to the received device capability information.

[00126] The embodiments disclosed herein can be implemented through at least one software program running on at least one hardware device and performing network management functions to control the elements. The elements can be at least one of a hardware device or a combination of hardware devices and software modules.

[00127] While specific language has been used to describe the disclosure, any limitations arising on account of the same are not intended. As would be apparent to a person in the art, various working modifications may be made to the method in order to implement the concept as taught herein.

[00128] The drawings and the forgoing description give examples of embodiments. Those skilled in the art will appreciate that one or more of the described elements may well be combined into a single functional element. Alternatively, certain elements may be split into multiple functional elements. Elements from one embodiment may be added to another embodiment. For example, orders of processes described herein may be changed and are not limited to the manner described herein.

[00129] Moreover, the actions of any flow diagram need not be implemented in the order shown; nor do all of the acts necessarily need to be performed. Also, those acts that are not dependent on other acts may be performed in parallel with the other acts. The scope of embodiments is by no means limited by these specific examples. Numerous variations, whether explicitly given in the specification or not. such as differences in structure, dimension, and use of material, are possible. The scope of embodiments is at least as broad as given by the following claims.

[00130] Benefits, other advantages, and solutions to problems have been described above with regard to specific embodiments. However, the benefits, advantages, solutions to problems, and any component(s) that may cause any benefit, advantage, or solution to occur or become more pronounced are not to be construed as a critical, required, or essential feature or component of any or all the claims.

[00131] The foregoing description of the specific embodiments will so fully reveal the general nature of the embodiments herein that others can, by applying current knowledge, readily modify and/or adapt for various applications such specific embodiments without departing from the generic concept, and, therefore, such adaptations and modifications should and are intended to be comprehended within the meaning and range of equivalents of the disclosed embodiments. It is to be understood that the phraseology or terminology employed herein is for the purpose of description and not of limitation. Therefore, while the embodiments herein have been described in terms of at least one embodiment, those skilled in the art will recognize that the embodiments herein can be practiced with modification within the spirit and scope of the embodiments as described herein.

Claims

WE CLAIM:

1. A method comprising: receiving, by a gNB, one or more UE-based Artificial Intelligence/Machine Learning (AI/ML) model output parameters and/or Key Performance Indicators (KPIs) and device capability information from a User Equipment (UE) for at least one selected functionality or use-case; comparing, by the gNB, the one or more UE-based AI/ML model output parameters and/or KPIs with corresponding predefined benchmarked values; and upon determining that at least one of the one or more UE-based AI/ML model output parameters and/or KPIs is below the corresponding predefined benchmarked value: configuring, by the gNB, the UE to: fallback to non-AI/ML-based operations for the at least one selected functionality or use-case; and one of disable or continue AI/ML-based measurements and/or predictions for the at least one selected functionality or use-case based on one or more predefined criteria corresponding to the received device capability information.

2. The method as claimed in claim 1, wherein configuring the UE to one of disable or continue AI/ML-based measurements and/or prediction comprises: determining, by the gNB, a radio resource overhead at the gNB, an air interface overhead, resource consumption at the UE and the gNB, and AI/ML-based capabilities of the UE; analyzing, by the gNB, the determined radio resource overhead at the gNB, air interface overhead, resource consumption at the UE and the gNB, and AI/ML-based capabilities of the UE, and deriving a priority' of the AI/ML-based measurements and/or predictions for the UE; in response to determining that the derived priority is above a predetermined threshold value, configuring, by the gNB, the UE to continue the AI/ML-based measurements and/or predictions for the at least one selected functionality or use-case; and in response to determining that the derived priority is below the predetermined threshold value, configuring, by the gNB, the UE to disable the AI/ML-based measurements and/or predictions for the at least one selected functionality or use-case.

3. The method as claimed in claim 2, further comprising: in response to configuring the UE to continue the AI/ML-based measurements and/or predictions for the at least one selected functionality or use-case, receiving, by the gNB, from the UE, an indication message, based on one or more pre-determined KPIs or parameters satisfying pre-determined threshold values, to initiate re-configuring the AI/ML based measurements and/or predictions for the at least one selected functionality or use-case and disable the non-AI/ML based operations.

4. The method as claimed in claim 1, further comprising: in response to configuring the UE to disable the AI/ML-based measurements and/or predictions for the at least one selected functionality' or use-case, monitoring, by the gNB, one or more pre-determined non- AI/ML-based KPIs/parameters; upon determining that at least one of the one or more pre-determined non- AI/ML-based KPIs/parameters is below a corresponding non-AI/ML-based predefined threshold value: enabling, by the gNB, the AI/ML-based measurements and/or predictions for the at least one selected functionality' or use-case; and disabling the non-AI/ML-based operations.

5. The method as claimed in claim 1, further comprising: configuring, by the gNB, the UE to indicate one of the AI/ML-based measurements and/or predictions or the non-AI/ML-based operations for the at least one selected functionality or use-case using a 1 -bit FLAG in at least one of a data collection report, a measurement report, or an AI/ML output delivery message.

6. The method as claimed in claim 1, wherein receiving comprises receiving a Channel State Information (CSI) feedback comprising at least one of the one or more UE-based AI/ML model output parameters and/or KPIs or the device capability information.

7. The method as claimed in claim 1, wherein the at least one selected functionality or usecase comprises one or more of channel state information (CSI) feedback, beam management, and positioning-related use-cases.

8. The method as claimed in claim 1, wherein the one or more predefined criteria includes a correlation between the received device capability infomiation and one or more air interface overhead, signaling overhead. UE power consumption, UE subscription type, UE and gNB processing overhead, and additional radio resource overhead caused by the AI/ML based measurements and/or predictions.

9. The method as claimed in claim 1, wherein the method is performed using at least one of a L2 and a L3 message.

10. An apparatus configured to: receive one or more UE-based Artificial Intelligence/Machine Learning (AI/ML) model output parameters and/or Key Performance Indicators (KPIs) and device capability information from a User Equipment (UE) for at least one selected functionality or use-case; compare the one or more UE-based AI/ML model output parameters and/or KPIs with corresponding predefined benchmarked values; and upon determining that at least one of the one or more UE-based AI/ML model output parameters and/or KPIs is below the corresponding predefined benchmarked value: configure the UE to: fallback to non-AI/ML-based operations for the at least one selected functionality or use-case; and one of disable or continue AI/ML-based measurements and/or predictions for the at least one selected functionality or use-case based on one or more predefined criteria corresponding to the received device capability infonnation.

1 1 . The apparatus as claimed in claim 10, wherein to configure the UE to perform one of disable or continue AI/ML-based measurements and/or prediction, the apparatus is configured to: determine a radio resource overhead at a gNB, an air interface overhead, resource consumption at the UE and the gNB, and AI/ML-based capabilities of the UE; analyze the determined radio resource overhead at the gNB, air interface overhead, resource consumption at the UE and the gNB, and AI/ML-based capabilities of the UE, and deriving a priority of the AI/ML-based measurements and/or predictions for the UE; in response to determining that the derived priority is above a predetermined threshold value, configure the UE to continue the AI/ML-based measurements and/or predictions for the at least one selected functionality or use-case; and in response to determining that the derived priority is below the predetermined threshold value, configure the UE to disable the AI/ML-based measurements and/or predictions for the at least one selected functionality⁷ or use-case.

12. The apparatus as claimed in claim 11, further configured to: in response to configuring the UE to continue the AI/ML-based measurements and/or predictions for the at least one selected functionality or use-case, receive from the UE, an indication message, based on one or more pre-determined KPIs or parameters satisfying predetermined threshold values, to initiate re-configuring the AI/ML based measurements and/or predictions for the at least one selected functionality or use-case and disable the non-AI/ML based operations.

13. The apparatus as claimed in claim 10, further configured to: in response to configuring the UE to disable the AI/ML-based measurements and/or predictions for the at least one selected functionality or use-case, monitor one or more predetermined non-AI/ML-based KPIs/parameters; upon determining that at least one of the one or more pre-determined non-AI/ML-based KPIs/parameters is below a corresponding non-AI/ML-based predefined threshold value: enable the AI/ML-based measurements and/or predictions for the at least one selected functionality or use-case; and disable the non-AI/ML-based operations.

14. The apparatus as claimed in claim 10, further configured to: configure the UE to indicate one of the AI/ML-based measurements and/or predictions or the non-AI/ML-based operations for the at least one selected functionality or use-case using a 1 -bit FLAG in at least one of a data collection report, a measurement report message, or a AI/ML output delivery message.

15. The apparatus as claimed in claim 10, further configured to: receive a channel state information (CSI) feedback comprising at least one of the one or more UE-based AI/ML model output parameters and/or KPIs or the device capability information.

16. The apparatus as claimed in claim 10, wherein the at least one selected functionality or use-case comprises one or more of channel state information (CSI) feedback, beam management, and positioning-related use-cases.

17. The apparatus as claimed in claim 10, wherein the one or more predefined criteria includes a correlation between the received device capability information and one or more air interface overhead, signaling overhead, UE power consumption, UE subscription ty pe, UE and gNB processing overhead, and additional radio resource overhead caused by the AI/ML based measurements and/or predictions.

18. A non-transitory computer-readable medium storing instructions, the instructions comprising: one or more instructions that, when executed by a network device comprising one or more processors, cause the one or more processors to: receive one or more UE-based Artificial Intelligence/Machine Learning (AI/ML) model output parameters and/or Key Performance Indicators (KPIs) and device capability information from a User Equipment (UE) for at least one selected functionality or use-case; compare the one or more UE-based AI/ML model output parameters and/or KPIs with corresponding predefined benchmarked values; and upon determining that at least one of the one or more UE-based AI/ML model output parameters and/or KPIs is below the corresponding predefined threshold value: configure the UE to: fallback to non-AI/ML-based operations for the at least one selected functionality or use-case; and one of disable or continue AI/ML-based measurements and/or predictions for the at least one selected functionality or use-case based on one or more predefined criteria corresponding to the received device capability information.