# US 2026/0052075 A1 · Precoded reference signal for model monitoring for ml-based csi feedback

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2026/0052075 A1
- Document ID: US20260052075A1
- Status: Published application
- Kind: A1
- Application number: 19/103,816
- Filed: 2022-11-04
- Published: 2026-02-19
- Earliest claimed: 2022-11-04
- Assignees: QUALCOMM Incorporated
- Applicants: QUALCOMM Incorporated
- Inventors: Jay Kumar SUNDARARAJAN; Chenxi HAO; June NAMGOONG; Taesang YOO
- Web page: https://patentgrep.com/patent/US20260052075A1

## CPC

- G06N 3/045: Combinations of networks
- H04L 41/16: using machine learning or artificial intelligence
- H04B 7/0421: {utilizing implicit feedback, e.g. steered pilot signals}
- H04B 7/0626: {Channel coefficients, e.g. channel state information [CSI]}
- H04L 25/0202: {Channel estimation}
- G06N 3/08: Learning methods
- G06N 3/084: Backpropagation, e.g. using gradient descent
- H04L 25/0224: {using sounding signals}
- H04L 25/0254: {using neural network algorithms}
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- H04L: TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION (arrangements common to telegraphic and telephonic communication H04M)
- H04B: TRANSMISSION

## Abstract

A method for wireless communication at a user equipment (UE) and related apparatus are provided. In the method, the UE receives a precoded pilot signal from a network entity. The precoded pilot signal includes a pilot signal that is precoded with a precoder based on a machine-learning (ML) model. The UE further reports an ML model performance monitoring result, including a comparison between the target channel estimation and the reconstructed channel estimation for a channel between the UE and the network entity based on the precoded pilot signal. The target channel estimation may be based on a target channel station information (CSI). The method allows the UE to monitor the performance of an ML model without the overhead of conveying the target CSI to the network entity. Hence, it improves the efficiency and reliability of wireless communication.

## Claims

### Claim 1

An apparatus for wireless communication at a user equipment (UE), comprising:
memory; and
at least one processor coupled to the memory and, based at least in part on information stored in the memory, the at least one processor is configured to:
receive, from a network entity, a precoded pilot signal, the precoded pilot signal comprising a pilot signal that is precoded with a machine-learning (ML) model based precoder; and
report an ML model performance monitoring result comprising a comparison between a target channel estimation and a reconstructed channel estimation for a channel between the UE and the network entity based on the precoded pilot signal.

### Claim 2

(canceled)

### Claim 3 (depends on 1)

The apparatus of claim 1, wherein the ML model based precoder is based on reconstructed channel station information (CSI) for the channel between the UE and the network entity.

### Claim 4

(canceled)

### Claim 5 (depends on 3)

The apparatus of claim 3, wherein the reconstructed channel estimation is based on a channel matrix and the reconstructed CSI, and the target channel estimation is based on the channel matrix and a target CSI.

### Claim 6 (depends on 5)

The apparatus of claim 5, wherein the at least one processor is further configured to:
receive, from the network entity, a non-precoded pilot signal comprising the pilot signal without precoding with the ML model based precoder, wherein the channel matrix is based on the non-precoded pilot signal.

### Claim 7 (depends on 5)

The apparatus of claim 5, wherein the at least one processor is further configured to:
determine the channel matrix prior to receiving the precoded pilot signal.

### Claim 8 (depends on 1)

The apparatus of claim 1, wherein the at least one processor is further configured to:
estimate, based on the precoded pilot signal, the reconstructed channel estimation for the channel between the UE and the network entity; and
monitor a performance of the ML model by comparing the reconstructed channel estimation with the target channel estimation of the channel.

### Claim 9 (depends on 8)

The apparatus of claim 8, wherein to monitor the performance of the ML model, the at least one processor is configured to:
compute a first signal quality metric associated with a reconstructed CSI; and
compute a second signal quality metric associated with a target CSI, wherein the comparison reported to the network entity includes the comparison of the first signal quality metric and the second signal quality metric.

### Claim 10 (depends on 9)

The apparatus of claim 9, wherein the ML model performance monitoring result indicates an ML model failure based on a difference between the first signal quality metric and the second signal quality metric being greater than a quality threshold.

### Claim 11 (depends on 10)

The apparatus of claim 10, wherein the at least one processor is further configured to:
update or switch the ML model in response to the ML model failure.

### Claim 12 (depends on 1)

The apparatus of claim 1, wherein the at least one processor is further configured to:
transmit, to the network entity, a CSI feedback message, wherein the pilot signal is precoded based on reconstructed CSI reconstructed by the ML model based on the CSI feedback message.

### Claim 13

(canceled)

### Claim 14

An apparatus for wireless communication at a network entity, comprising:
memory; and
at least one processor coupled to the memory and, based at least in part on information stored in the memory, the at least one processor is configured to:
transmit, to a user equipment (UE), a precoded pilot signal comprising a pilot signal precoded with a precoder based on a machine-learning (ML) model; and
receive, from the UE, an ML model performance monitoring result comprising a comparison between a target channel estimation and a reconstructed channel estimation for a channel between the UE and the network entity based on the precoded pilot signal.

### Claim 15

(canceled)

### Claim 16 (depends on 14)

The apparatus of claim 14, wherein the at least one processor is further configured to:
receive, from the UE, a channel state information (CSI) feedback message for the channel between the UE and the network entity; and
precode the pilot signal based on the ML model at the network entity and reconstructed CSI for the channel between the UE and the network entity, wherein the reconstructed CSI is generated by the ML model based on the CSI feedback message.

### Claim 17

(canceled)

### Claim 18 (depends on 16)

The apparatus of claim 16, wherein the reconstructed channel estimation is based on a channel matrix and a reconstruction of the CSI feedback message, and the target channel estimation is based on the channel matrix and a target CSI.

### Claim 19 (depends on 18)

The apparatus of claim 18, wherein the at least one processor is further configured to:
transmit, to the UE, a non-precoded pilot signal comprising the pilot signal without the precoder, wherein the channel matrix is based on the non-precoded pilot signal.

### Claim 20 (depends on 18)

The apparatus of claim 18, wherein the channel matrix is determined prior to transmitting the precoded pilot signal.

### Claim 21 (depends on 18)

The apparatus of claim 18, wherein the ML model performance monitoring result comprises the comparison between a first signal quality metric associated with the reconstructed CSI and a second signal quality metric associated with the target CSI.

### Claim 22 (depends on 21)

The apparatus of claim 21, wherein the ML model performance monitoring result indicates an ML model failure based on a difference between the first signal quality metric and the second signal quality metric being greater than a quality threshold.

### Claim 23 (depends on 22)

The apparatus of claim 22, wherein the at least one processor is further configured to:
update or switch the ML model based on the ML model performance monitoring result.

### Claim 24

(canceled)

### Claim 25

(canceled)

### Claim 26

A method for wireless communication at a user equipment (UE) comprising:
receiving, from a network entity, a precoded pilot signal, the precoded pilot signal comprising a pilot signal that is precoded with a machine-learning (ML) model based precoder; and
reporting an ML model performance monitoring result comprising a comparison between a target channel estimation and a reconstructed channel estimation for a channel between the UE and the network entity based on the precoded pilot signal.

### Claim 27 (depends on 26)

The method of claim 26, wherein the ML model based precoder is based on reconstructed channel station information (CSI) for the channel between the UE and the network entity.

### Claim 28

(canceled)

### Claim 29

(canceled)

### Claim 30

(canceled)