Search

KR-20260066785-A - Label generation method for online learning of channel sounding using pseudo-random data

KR20260066785AKR 20260066785 AKR20260066785 AKR 20260066785AKR-20260066785-A

Abstract

The radio transceiver unit (WTRU) may receive pseudo-random data (PRD), one or more channel state information reference signals (CSI-RS), and/or one or more learning reference signals (L-RS). The WTRU may determine a CSI measurement based on the received CSI-RS and/or determine an L-RS measurement based on the received L-RS. The WTRU may generate one or more CSI labels based on the PRD, CSI-RS measurements, and/or L-RS measurements. The WTRU may generate one or more output CSIs using an artificial intelligence/machine learning (AI/ML) model and based on the PRD and/or CSI-RS measurements. The WTRU may train an AI/ML model with one or more CSI labels and one or more output CSIs.

Inventors

  • 아르파오이 모하메드 아미네
  • 피에트라스키 필립
  • 페리아드 프랑수아
  • 장 구오동

Assignees

  • 인터디지탈 패튼 홀딩스, 인크

Dates

Publication Date
20260512
Application Date
20240919
Priority Date
20230921

Claims (20)

  1. In a method implemented by a wireless transmit/receive unit (WTRU), A step of receiving pseudo-random data (PRD); A step of receiving one or more channel state information reference signals (CSI-RS); A step of receiving one or more learning reference signals (L-RS); A step of determining a CSI measurement based on the received CSI-RS; A step of determining an L-RS measurement based on the received L-RS; A step of generating one or more CSI labels based on the above PRD, the above CSI-RS measurement, and the above L-RS measurement; A step of generating one or more output CSIs using an artificial intelligence/machine learning (AI/ML) model and based on the PRD and CSI-RS measurements; and Step of training the AI/ML model with the above one or more CSI labels and the above one or more output CSIs A method implemented by WTRU that includes
  2. In claim 1, A step of determining a change in the loss function of the AI/ML model based on the one or more CSI labels and the one or more output CSIs; and A step of determining whether the training is complete based on a comparison between the value of the loss function and a threshold value. A method implemented by WTRU that further includes.
  3. In claim 2, A method implemented by WTRU in which the loss function of the above AI/ML model determines a metric for training accuracy during the training of the above AI/ML model.
  4. In claim 1, Step to determine model drift metrics; A step of determining the performance of the AI/ML model for CSI estimation based on model drift by applying a statistical test to the above CSI-RS measurement; and Step of retraining the AI/ML model based on the above model drift metric exceeding the drift threshold A method implemented by WTRU that further includes.
  5. In claim 4, A step of transmitting a report to a network in response to an indication that the above model drift metric exceeds the above drift threshold. A method implemented by WTRU, which further includes, wherein the report includes details of the model drift or recommendations for tuning the AI/ML model.
  6. In claim 1, A step of transmitting to a network a request to be composed of a report based on the above CSI-RS measurement and one or more training assignments; and A step of receiving from the network an indication of whether the request to be composed of one or more training assignments has been approved. A method implemented by WTRU that further includes.
  7. In claim 6, The above indication is a method implemented by WTRU that further includes seed information used to generate the above PRD.
  8. In claim 6, Step of receiving PRD training allocation configuration It further includes, and the above PRD training assignment indicates a PRD training assignment of type 1, a PRD training assignment of type 2, or a PRD training assignment of type 3; The above-mentioned first type PRD training assignment configures the WTRU to use the PRD and the CSI-RS measurements for a first resource block set carrying the PRD and the CSI-RS; The above-mentioned second type of PRD training assignment configures the WTRU to use the PRD and the CSI-RS measurements for a second resource block set containing data that does not carry the PRD and a resource element containing data that does not carry the PRD and the PRD, wherein the resource element containing data that does not carry the PRD and the PRD occupy the same resource block within the second resource block set; A method implemented by WTRU in which the above-mentioned third type of PRD training assignment configures the WTRU to use the PRD and the CSI-RS measurements for a third resource block set containing data that does not carry the PRD and a resource element containing data that does not carry the PRD and the PRD, wherein the resource element containing data that does not carry the PRD and the PRD occupy different but overlapping resource blocks within the third resource block set.
  9. In claim 8, The above-mentioned first, second, and third types of PRD resource allocation is a method implemented by WTRU, which is signaled semi-statically through a system information block or through a broadcast message in downlink control information (DCI).
  10. In claim 1, A step of transmitting a first message including an online training request - said online training request indicates the number of resource blocks carrying PRDs for training, channel statistics measurements, a channel diversification request, and the priority of the request -; and Step of receiving a second message containing an online training response A method implemented by WTRU, further comprising, wherein the online training response indicates the location of the resource block including the PRD, the allocated online training time duration, the allocated online training start time, and the allocated online training L-RS.
  11. In a wireless transceiver unit (WTRU) including a processor and memory, the processor and memory are: Receive pseudo-random data (PRD); Receive one or more Channel State Information Reference Signals (CSI-RS); Receive one or more learning reference signals (L-RS); Determining CSI measurements based on the received CSI-RS above; Determining an L-RS measurement based on the received L-RS above; Generate one or more CSI labels based on the above PRD, the above CSI-RS measurement, and the above L-RS measurement; Generating one or more output CSIs using an artificial intelligence/machine learning (AI/ML) model and based on the PRD and CSI-RS measurements; WTRU configured to train the AI/ML model with the above one or more CSI labels and the above one or more output CSIs.
  12. In claim 11, the processor also: Determining a change in the loss function of the AI/ML model based on the above one or more CSI labels and the above one or more output CSIs; WTRU configured to determine whether the training is complete based on a comparison between the value of the loss function and a threshold value.
  13. In claim 12, The loss function of the above AI/ML model is WTRU, which determines a metric for training accuracy during the training of the above AI/ML model.
  14. In claim 11, the processor also: Determine the model drift metric; Determining the performance of the AI/ML model for CSI estimation based on model drift by applying a statistical test to the above CSI-RS measurement; WTRU configured to retrain the AI/ML model based on the above model drift metric exceeding a drift threshold.
  15. In claim 14, the processor also: WTRU configured to send a report to a network in response to an indication that the model drift metric exceeds the drift threshold, wherein the report includes details of the model drift or recommendations for tuning the AI/ML model.
  16. In claim 1, the processor also: Sending a request to the network consisting of a report based on the above CSI-RS measurement and one or more training assignments; WTRU configured to receive from the network an indication of whether the request to be composed of one or more training assignments has been approved.
  17. In claim 16, The above indication is a WTRU that further includes seed information used to generate the above PRD.
  18. In claim 16, the processor also: Configured to receive a PRD training assignment configuration, wherein the PRD training assignment indicates a PRD training assignment of type 1, a PRD training assignment of type 2, or a PRD training assignment of type 3; The above-mentioned first type PRD training assignment configures the WTRU to use the PRD and the CSI-RS measurements for a first resource block set carrying the PRD and the CSI-RS; The above-mentioned second type of PRD training assignment configures the WTRU to use the PRD and the CSI-RS measurements for a second resource block set containing data that does not carry the PRD and a resource element containing data that does not carry the PRD and the PRD, wherein the resource element containing data that does not carry the PRD and the PRD occupy the same resource block within the second resource block set; The above-mentioned third type of PRD training assignment configures the WTRU to use the PRD and the CSI-RS measurements for a third resource block set containing data that does not carry the PRD and a resource element containing data that does not carry the PRD and the PRD, wherein the resource element containing data that does not carry the PRD and the PRD occupy different but overlapping resource blocks within the third resource block set.
  19. In claim 18, The above-mentioned first, second, and third types of PRD resource allocations are signaled semi-statically through system information blocks or via broadcast messages in downlink control information (DCI), WTRU.
  20. In claim 11, the processor also: Transmitting a first message including an online training request - said online training request indicates the number of resource blocks carrying PRD for training, channel statistics measurements, a channel diversification request, and the priority of the request -; WTRU configured to receive a second message including an online training response, wherein the online training response indicates the location of the resource block including the PRD, the allocated online training time duration, the allocated online training start time, and the allocated online training L-RS.

Description

Label generation method for online learning of channel sounding using pseudo-random data This application claims the benefit of U.S. Provisional Patent Application No. 63/584,204 filed September 21, 2023, the entire contents of which are incorporated herein by reference. Artificial intelligence (AI) can be broadly defined as behavior exhibited by machines that perceive, reason, adapt, and/or act by mimicking cognitive functions. AI components may refer to the realization of behavior and/or compliance with requirements by learning based on data without the explicit configuration of a sequence of steps of actions. Such AI components can enable the learning of complex behaviors that might be difficult to specify and/or implement when using legacy methods. Machine learning (ML) can refer to a type of algorithm that solves problems based on learning through experience (e.g., data) without being explicitly programmed (e.g., constructing a rule set). ML can be considered a subset of AI. Different ML paradigms can be conceived based on the nature of the data and/or feedback available to the learning algorithm. In one example, a supervised learning approach may involve learning a function that maps an input to an output based on labeled training examples. Each training example may be a pair consisting of, for example, an input and its corresponding output. In another example, an unsupervised learning approach may involve detecting patterns in data without existing labels. In yet another example, a reinforcement learning approach may involve performing a sequence of actions in an environment to maximize cumulative rewards. A wireless transmit/receive unit (WTRU) can receive pseudo-random data (PRD), one or more channel state information reference signals (CSI-RS), and/or one or more learning reference signals (L-RS). The WTRU can determine a CSI measurement based on the received CSI-RS measurement. The WTRU can determine an L-RS measurement based on the received L-RS. The WTRU can generate one or more CSI labels based on the PRD, CSI-RS measurement, and/or L-RS measurement. The WTRU can generate one or more output CSIs using an artificial intelligence/machine learning (AI/ML) model and based on the PRD and/or CSI-RS measurement. The WTRU can train an AI/ML model with one or more CSI labels and one or more output CSIs. WTRU can determine changes to the loss function of an AI/ML model based on one or more CSI labels and one or more output CSIs. WTRU can determine whether training is complete based on a comparison of the loss function value with a threshold. The loss function of an AI/ML model can determine metrics for training accuracy during the training of the AI/ML model. WTRU can determine the model drift metric. WTRU can determine the performance of the AI/ML model for CSI estimation based on model drift by applying a statistical test to the CSI-RS measurement. WTRU can retrain the AI/ML model based on whether the model drift metric exceeds a drift threshold. WTRU can send a report to the network in response to an indication that the model drift metric exceeds a threshold. The report may include details of the model drift or recommendations for tuning the AI/ML model. The WTRU may send a request to the network to consist of a report based on CSI-RS measurements and/or one or more training assignments. The WTRU may receive from the network an indication (e.g., a response) of whether the request to consist of one or more training assignments has been approved. The indication (e.g., a response) may further include seed information used to generate the PRD. A WTRU can receive a PRD training assignment configuration. A PRD training assignment may represent a PRD training assignment of type 1, a PRD training assignment of type 2, and/or a PRD training assignment of type 3. A PRD training assignment of type 1 may configure a WTRU to use PRD and CSI-RS measurements for a first set of resource blocks carrying PRD and CSI-RS. A PRD training assignment of type 2 may configure a WTRU to use PRD and/or CSI-RS measurements for a second set of resource blocks carrying resource elements containing data that do not carry PRD, CSI-RS, and/or PRD. A PRD and a resource element containing data that does not carry PRD may occupy the same resource block. A PRD training assignment of type 3 may configure a WTRU to use PRD and CSI-RS measurements for a third set of resource blocks carrying resource elements containing data that does not carry PRD, CSI-RS, and/or PRD. Resource elements containing PRD and data that do not carry PRD can occupy different but overlapping resource blocks. The allocation of PRD resources of the first, second, and/or third types may be signaled semi-statically through system information blocks and/or via broadcast messages in downlink control information (DCI). The WTRU may transmit a first message containing an online training request. The online training request may indicate the number of resource blocks carryi