Search

KR-20260062796-A - METHOD AND APPARATUS FOR AI BASED CHANNEL ESTIAMTION IN WIRELESS COMMUNICATION SYSTEM

KR20260062796AKR 20260062796 AKR20260062796 AKR 20260062796AKR-20260062796-A

Abstract

The present disclosure relates to an AI-based channel estimation method and apparatus in a wireless communication system. A method performed in a communication device that performs channel estimation in a wireless communication system according to an embodiment of the present disclosure comprises: a process of obtaining initial channel estimation information including channel noise by performing channel estimation using a reference signal; a process of determining whether to perform an AI-based channel noise removal operation; and a process of obtaining channel estimation information with the channel noise removed by performing the AI-based channel removal operation. The AI-based channel removal operation comprises: a process of performing data-assisted channel estimation using data detected in a data region adjacent to the reference signal in a data channel as a virtual reference signal; a process of obtaining AI model parameters for fine-tuning an AI model for channel noise removal based on the results of the data-assisted channel estimation; and a process of outputting channel estimation information with the channel noise removed using an AI model fine-tuned based on the obtained AI model parameters.

Inventors

  • 이익범
  • 전요셉
  • 하성영
  • 양하영
  • 양경철

Assignees

  • 삼성전자주식회사
  • 포항공과대학교 산학협력단

Dates

Publication Date
20260507
Application Date
20250520
Priority Date
20241029

Claims (16)

  1. A method performed in a communication device that performs channel estimation in a wireless communication system, A process of obtaining initial channel estimation information containing channel noise by performing channel estimation using a reference signal; A process for determining whether to perform AI-based channel noise reduction; The method includes a process of obtaining channel estimation information from which channel noise has been removed by performing the above AI-based channel removal operation, The above AI-based channel removal operation is, A process of performing data support channel estimation using data detected in a data region adjacent to the reference signal in a data channel as a virtual reference signal; A process of obtaining AI model parameters for fine-tuning an AI (artificial intelligence) model for channel noise removal based on data-assisted channel estimation results; and A method comprising the process of outputting channel estimation information with channel noise removed using an AI model finely tuned based on the AI model parameters obtained above.
  2. Since Article 1 exists, The above AI model for removing channel noise uses a neural network (NN)-based model.
  3. Since Article 1 exists, The above reference signal is a demodulation reference signal (DM-RS) placed at a fixed location within the resource block (RB), and The process of performing the above data support channel estimation is, A method further comprising the process of performing channel estimation using the above DM-RS and the virtual DM-RS, which is the virtual reference signal.
  4. Since Article 1 exists, The above communication device is a base station, and The process of acquiring AI model parameters for the above fine-tuning is, The process of the above base station acquiring first training samples from a plurality of terminals within a cell; and A method further comprising the process of obtaining gradient-adjusted AI model parameters using the first training samples above.
  5. Since there is Article 4, The above first training samples are a method of using online data obtained during communication from the above multiple terminals at the current time.
  6. Since there is Article 4, The above communication device is a base station, and The above AI model prior to fine-tuning utilizes a method in which a pre-trained AI model is used based on second training samples obtained from channel estimation information when the base station communicates with multiple terminals at a past time.
  7. Since there is Article 6, A method in which the amount of data of the first training samples is less than the amount of data of the second training samples.
  8. Since there is Article 6, A method further comprising the process of pre-training the AI model through meta-learning using the second training samples.
  9. In a communication device that performs channel estimation in a wireless communication system, Transmitter/Receiver; One or more processors including processing circuitry; and The communication device includes a memory for storing instructions, and when the instructions are executed individually or collectively by one or more processors, the communication device Initial channel estimation information including channel noise is obtained by performing channel estimation using a reference signal, and Determine whether to perform AI-based channel noise reduction, and The above AI-based channel removal operation is performed to obtain channel estimation information from which the channel noise has been removed, and For the above AI-based channel removal operation, the above commands are for the communication device at least, Data support channel estimation is performed using data detected in a data region adjacent to the reference signal in the data channel as a virtual reference signal, and Based on the data-assisted channel estimation results, AI model parameters for fine-tuning the AI (artificial intelligence) model for channel noise removal are obtained, and A communication device that causes to output channel estimation information with channel noise removed using an AI model finely tuned based on the AI model parameters obtained above.
  10. Since there is Article 9, The above AI model for channel noise removal is a communication device that uses a neural network (NN)-based model.
  11. Since there is Article 9, The above reference signal is a demodulation reference signal (DM-RS) placed at a fixed location within the resource block (RB), and When the above commands are executed individually or collectively by the one or more processors, the communication device, A communication device that causes channel estimation to be performed using the above DM-RS and the virtual DM-RS, which is the virtual reference signal.
  12. Since there is Article 9, The above communication device is a base station, and When the above commands are executed individually or collectively by the one or more processors, the communication device, Acquiring first training samples from a plurality of terminals within a cell of the above-mentioned base station, and A communication device that causes to obtain gradient-adjusted AI model parameters using the above-mentioned first training samples.
  13. Since there is Article 12, The above first training samples are a communication device that uses online data obtained during communication from the above multiple terminals at the current time.
  14. Since there is Article 12, The above communication device is a base station, and The above AI model prior to fine-tuning is a communication device that uses a pre-trained AI model using second training samples obtained based on channel estimation information when the base station communicates with multiple terminals at a past time.
  15. Since there is Article 14, A communication device in which the amount of data of the first training samples is less than the amount of data of the second training samples.
  16. Since there is Article 14, When the above commands are executed individually or collectively by the one or more processors, the communication device, A communication device that causes the AI model to be pre-trained through meta-learning using the second training samples.

Description

Method and apparatus for AI-based channel estimation in wireless communication system The present disclosure relates to a method and apparatus for estimating a channel in a wireless communication system. To meet the increasing demand for wireless data traffic following the 4G system (i.e., LTE (long-term evolution) system), the 5G system (i.e., NR (New Radio) system) has been developed and commercialized. The 5G system can be implemented in the mmWave band. To mitigate path loss and increase the transmission distance of radio waves in the mmWave band, beamforming, massive array multiple-input multiple-output (massive MIMO), full-dimensional multiple-input multiple-output (Full Dimensional MIMO: FD-MIMO), array antenna, analog beamforming, and large-scale antenna technologies are being discussed for 5G systems. Wireless communication systems such as 4G and 5G systems utilize the Orthogonal Frequency Division Multiplexing (OFDM) method, which is robust against interference between signals and frequency selective fading and is suitable for broadband signal transmission. Frequency selective fading is a type of fading (signal attenuation) phenomenon that occurs in a multipath communication environment, referring to a phenomenon where different frequency components attenuate to different degrees, causing the signal to be strongly affected only in specific frequency bands. Frequency selective fading primarily occurs in communication environments with severe delay spread caused by multipath, and due to this frequency selectivity, the transmitted signal in a wireless communication system may be weak in some frequency bands within the entire frequency band while being received strongly in other frequency bands. In OFDM systems, overall performance is directly influenced by channel estimation performance in the frequency domain; therefore, various algorithms are being researched to enhance the overall performance of communication systems by efficiently performing channel estimation in the frequency domain. In wireless communication systems, improving channel estimation performance can lead to increased cell coverage and throughput. Multiple-Input Multiple-Output Orthogonal Frequency Division Multiplexing (MIMO-OFDM) systems, which are OFDM-based communication systems supporting MIMO—a multi-antenna technology—are a technology capable of dramatically improving the transmission speed and stability of wireless communication and are currently utilized as a core element of commercially available 5G NR (New Radio) standards. By simultaneously transmitting multiple data streams through multiple antennas, MIMO-OFDM systems can secure channel diversity and increase data transmission rates and channel capacity. Furthermore, MIMO-OFDM systems can effectively mitigate inter-symbol interference occurring in multipath channels. However, since wireless channels possess complex characteristics that vary over time and space, failure to perform accurate channel estimation in MIMO-OFDM systems that utilize wireless channels for communication can negatively impact the performance of the communication system. Furthermore, channel delay and the Doppler effect in MIMO-OFDM systems can cause severe distortion, leading to data loss and an increase in the error rate. Therefore, the need for more advanced channel estimation technology is essential to improve the performance of MIMO-OFDM systems. FIG. 1 is a diagram showing an example of the basic structure of time-frequency resources of a 5G system. FIG. 2 is a diagram showing an example of a frame, subframe, and slot structure of a 5G system. FIG. 3 is a diagram showing an example of a bandwidth portion setting in a 5G system. FIG. 4 is a diagram briefly showing the configuration of a MIMO-OFDM system as an example of a wireless communication system to which the present disclosure applies. FIG. 5 is a diagram showing an example of an AI-based channel noise removal method for channel estimation in a wireless communication system according to an embodiment of the present disclosure. FIG. 6 is a diagram showing an example of a configuration of a receiving device that performs AI-based channel noise removal for channel estimation in a wireless communication system according to an embodiment of the present disclosure. FIG. 7 is a drawing showing an example of a virtual DM-RS according to an embodiment of the present disclosure, FIG. 8 is a diagram illustrating an example of a scenario for performing meta-training and fine-tuning for AI-based channel noise removal in a wireless communication system according to an embodiment of the present disclosure. FIG. 9 is a diagram showing an example of a channel noise removal process using an AI model in a wireless communication system according to an embodiment of the present disclosure. FIG. 10 is a diagram showing an example of an AI-based channel noise removal method for channel estimation in a wireless communication system according to an embodiment of t