EP-4740562-A1 - METHOD AND APPARATUS OF MODEL MONITORING FOR ARTIFICIAL INTELLIGENCE (AI) -BASED CHANNEL STATE INFORMATION (CSI) PREDICTION

Abstract

In an aspect of the disclosure, a method, a computer-readable medium, and an apparatus are provided. The method may be performed by a UE. In certain configurations, the UE receives, from a base station, a signal to activate a model monitoring process for artificial intelligence (AI) /machine learning (ML) -based channel state information (CSI) prediction with a current model. The UE performs the AI/ML-based CSI prediction. In response to determining that an adjustment to the current model is required, the UE implements the adjustment to the current model. In certain configurations, calculation of performance metrics of the current model and monitoring of the current model may be performed at the UE or the base station.

Inventors

• HUANG, YU-CHING
• KYUNG, GYU BUM

Assignees

• MEDIATEK INC.

Dates

Publication Date

20260513

Application Date

20240626

Claims (20)

1. A method of wireless communication of a user equipment (UE) , comprising: receiving, from a base station, a signal to activate a model monitoring process for artificial intelligence (AI) /machine learning (ML) -based channel state information (CSI) prediction with a current model; performing the AI/ML-based CSI prediction; and in response to determining that an adjustment to the current model is required, implementing the adjustment to the current model.
2. The method of claim 1, further comprising: receiving, from the base station, configuration information for configuring CSI reference signal (CSI-RS) symbols for monitoring.
3. The method of claim 2, wherein: when a prediction interval is equal to an observation interval of the CSI-RS symbols, calculation of performance metrics of the current model is performed based on observed CSI-RS symbols and predicted CSI-RS symbols obtained in the AI/ML-based CSI prediction; and when the prediction interval is not equal to the observation interval of the CSI-RS symbols, calculation of the performance metrics of the current model is performed based solely on the predicted CSI-RS symbols.
4. The method of claim 1, further comprising: receiving, from the base station, a signal indicating the adjustment to the current model, wherein the base station is configured to perform calculation of performance metrics of the current model, and to monitor the current model for determining whether the adjustment to the current model is required.
5. The method of claim 4, further comprising: transmitting, to the base station, UE-specific assistance information including time domain channel property (TDCP) for assisting the base station in determining whether the adjustment to the current model is required.
6. The method of claim 4, further comprising: transmitting, to the base station, observed CSI reference signal (CSI-RS) and predicted CSI-RS, wherein the base station is configured to perform calculation of the performance metrics of the current model based on the observed CSI-RS and the predicted CSI-RS.
7. The method of claim 1, further comprising: performing calculation of performance metrics of the current model; determining whether the adjustment to the current model is required; and in response to determining that the adjustment to the current model is required, implementing the adjustment to the current model, and transmitting determination feedback of the adjustment to the current model to the base station.
8. The method of claim 7, further comprising: receiving, from the base station, UE-specific assistance information for assisting the UE in determining whether the adjustment to the current model is required.
9. The method of claim 1, further comprising: performing calculation of performance metrics of the current model; transmitting, to the base station, the performance metrics of the current model, wherein the base station is configured to determine whether the adjustment to the current model is required; and receiving, from the base station, a signal indicating the adjustment to the current model.
10. The method of claim 9, further comprising: transmitting, to the base station, UE-specific assistance information including time domain channel property (TDCP) for assisting the base station in determining whether the adjustment to the current model is required.
11. The method of claim 1, wherein the adjustment to the current model is: deactivating the current model, switching from the current model to a new model, performing fallback from the current model to a non-AI based model, or updating or fine-tuning the current model.
12. A method of wireless communication of a base station, comprising: transmitting, to a user equipment (UE) , a signal to activate a model monitoring process for artificial intelligence (AI) /machine learning (ML) -based channel state information (CSI) prediction with a current model, wherein the UE is configured to perform the AI/ML-based CSI prediction, and to implement an adjustment to the current model in response to determining that the adjustment is required.
13. The method of claim 12, further comprising: transmitting, to the UE, configuration information for configuring CSI reference signal (CSI-RS) symbols for monitoring.
14. The method of claim 13, wherein: when a prediction interval is equal to an observation interval of the CSI-RS symbols, calculation of performance metrics of the current model is performed based on observed CSI-RS symbols and predicted CSI-RS symbols obtained in the AI/ML-based CSI prediction; and when the prediction interval is not equal to the observation interval of the CSI-RS symbols, calculation of the performance metrics of the current model is performed based solely on the predicted CSI-RS symbols.
15. The method of claim 12, further comprising: performing calculation of performance metrics of the current model; monitoring the current model for determining whether the adjustment to the current model is required; and in response to determining that the adjustment to the current model is required, transmitting, to the UE, a signal indicating the adjustment to the current model.
16. The method of claim 15, further comprising: receiving, from the UE, UE-specific assistance information including time domain channel property (TDCP) for assisting the base station in determining whether the adjustment to the current model is required.
17. The method of claim 15, further comprising: receiving, from the UE, observed CSI reference signal (CSI-RS) and predicted CSI-RS.
18. The method of claim 12, further comprising: transmitting, to the UE, UE-specific assistance information including time domain channel property (TDCP) for assisting the UE in determining whether the adjustment to the current model is required; and receiving, from the UE, determination feedback of the adjustment to the current model to the base station.
19. The method of claim 12, further comprising: receiving, from the UE, performance metrics of the current model calculated by the UE; monitoring the current model for determining whether the adjustment to the current model is required; and in response to determining that the adjustment to the current model is required, transmitting, to the UE, a signal indicating the adjustment to the current model.
20. The method of claim 19, further comprising: receiving, from the UE, UE-specific assistance information including time domain channel property (TDCP) for assisting the base station in determining whether the adjustment to the current model is required.

Description

METHOD AND APPARATUS OF MODEL MONITORING FOR ARTIFICIAL INTELLIGENCE (AI) -BASED CHANNEL STATE INFORMATION (CSI) PREDICTION CROSS-REFERENCE TO RELATED APPLICATION (S) This application claims the benefits of U.S. Provisional Application Serial No. 63/511,693, entitled “Method and Apparatus of Model Monitoring for Artificial Intelligence (AI) -based Channel State Information (CSI) Prediction” and filed on July 3, 2023. The disclosure of the above-identified application is expressly incorporated by reference herein in their entirety. TECHNICAL FIELD The present disclosure relates generally to communication systems, and more particularly, to techniques of methods and apparatuses of model monitoring for artificial intelligence (AI) -based channel state information (CSI) prediction. BACKGROUND The statements in this section merely provide background information related to the present disclosure and may not constitute prior art. Wireless communication systems are widely deployed to provide various telecommunication services such as telephony, video, data, messaging, and broadcasts. Typical wireless communication systems may employ multiple-access technologies capable of supporting communication with multiple users by sharing available system resources. Examples of such multiple-access technologies include code division multiple access (CDMA) systems, time division multiple access (TDMA) systems, frequency division multiple access (FDMA) systems, orthogonal frequency division multiple access (OFDMA) systems, single-carrier frequency division multiple access (SC-FDMA) systems, and time division synchronous code division multiple access (TD-SCDMA) systems. These multiple access technologies have been adopted in various telecommunication standards to provide a common protocol that enables different wireless devices to communicate on a municipal, national, regional, and even global level. An example telecommunication standard is 5G New Radio (NR) . 5G NR is part of a continuous mobile broadband evolution promulgated by Third Generation Partnership Project (3GPP) to meet new requirements associated with latency, reliability, security, scalability (e.g., with Internet of Things (IoT) ) , and other requirements. Some aspects of 5G NR may be based on the 4G Long Term Evolution (LTE) standard. There exists a need for further improvements in 5G NR technology. These improvements may also be applicable to other multi-access technologies and the telecommunication standards that employ these technologies. SUMMARY The following presents a simplified summary of one or more aspects in order to provide a basic understanding of such aspects. This summary is not an extensive overview of all contemplated aspects, and is intended to neither identify key or critical elements of all aspects nor delineate the scope of any or all aspects. Its sole purpose is to present some concepts of one or more aspects in a simplified form as a prelude to the more detailed description that is presented later. In an aspect of the disclosure, a method, a computer-readable medium, and an apparatus are provided. The method may be performed by a UE. In certain configurations, the UE receives, from a base station, a signal to activate a model monitoring process for artificial intelligence (AI) /machine learning (ML) -based channel state information (CSI) prediction with a current model. The UE performs the AI/ML-based CSI prediction. In response to determining that an adjustment to the current model is required, the UE implements the adjustment to the current model. In another aspect of the disclosure, a method, a computer-readable medium, and an apparatus are provided. The method may be performed by a base station. In certain configurations, the base station transmits, to a UE, a signal to activate a model monitoring process for AI/ML-based CSI prediction with a current model. The UE is configured to perform the AI/ML-based CSI prediction, and to implement an adjustment to the current model in response to determining that the adjustment is required. To the accomplishment of the foregoing and related ends, the one or more aspects comprise the features hereinafter fully described and particularly pointed out in the claims. The following description and the annexed drawings set forth in detail certain illustrative features of the one or more aspects. These features are indicative, however, of but a few of the various ways in which the principles of various aspects may be employed, and this description is intended to include all such aspects and their equivalents. BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a diagram illustrating an example of a wireless communications system and an access network. FIG. 2 is a diagram illustrating a base station in communication with a UE in an access network. FIG. 3 illustrates an example logical architecture of a distributed access network. FIG. 4 illustrates an example physical architecture of a distributed access network. FIG