METHOD AND DEVICE FOR IMPROVING DATA RECEPTION PERFORMANCE IN COMMUNICATION SYSTEM

The present invention relates to a 6G communication system for achieving high data transmission rates and ultra-low latency beyond 4G and 5G communication systems. According to one embodiment of the present invention, provided are a method and device wherein, in order to improve the coverage of a tr...

Full description

Saved in:
Bibliographic Details
Main Authors PARK, Yosub, KIM, Suhwook, JANG, Hyeondeok, SEO, Bongsung
Format Patent
LanguageEnglish
French
Korean
Published 06.07.2023
Subjects
Online AccessGet full text

Cover

Loading…
More Information
Summary:The present invention relates to a 6G communication system for achieving high data transmission rates and ultra-low latency beyond 4G and 5G communication systems. According to one embodiment of the present invention, provided are a method and device wherein, in order to improve the coverage of a transmission end and the reception performance of a reception end, an AI model is trained on the PA input/output relationship at the reception end, and the trained AI model is used to compensate for PA nonlinearity in received data. In addition, according to one embodiment of the present invention, provided are a method and device wherein training of the AI model on the reception end is only instructed when training of the AI model is necessary, and a reference signal suitable for the training of the AI model is instructed. Accordingly, it is possible to reduce the signaling overhead, computational overhead, or power consumption generated during the training of the AI model. In addition, the reference signal having the same or similar characteristics as the data compensating for PA nonlinearity is used in the training of the AI model, and thus overfitting of the AI model can be prevented. Therefore, the AI model having optimal PA nonlinearity compensation performance can be obtained. La présente invention concerne un système de communication 6G permettant d'obtenir des vitesses de transmission de données élevées et une latence ultra-faible au-delà des systèmes de communication 4G et 5G. Un mode de réalisation de la présente invention concerne un procédé et un dispositif selon lesquels, afin d'améliorer la couverture d'une extrémité de transmission et les performances de réception d'une extrémité de réception, un modèle d'IA est appris sur la relation d'entrée/sortie PA au niveau de l'extrémité de réception, et le modèle AI appris est utilisé pour compenser la non-linéarité PA dans les données reçues. De plus, selon un mode de réalisation, l'invention concerne un procédé et un dispositif selon lesquels l'apprentissage du modèle d'IA sur l'extrémité de réception n'est demandé que lorsque l'apprentissage du modèle d'IA est nécessaire, et un signal de référence approprié à l'apprentissage du modèle d'IA est demandé. Par conséquent, il est possible de réduire le surdébit de signalisation, le surdébit de calcul ou la consommation d'énergie générée pendant l'apprentissage du modèle d'IA. De plus, le signal de référence ayant des caractéristiques identiques ou similaires à celles des données compensant la non-linéarité PA est utilisé dans l'apprentissage du modèle d'IA, ce qui permet d'éviter un surajustement du modèle d'IA. Il est donc possible d'obtenir un modèle d'IA ayant des performances optimales en matière de compensation de la non-linéarité PA. 본 개시는 4G, 5G 통신 시스템 이후 높은 데이터 전송 속도 및 초저지연시간을 달성하기 위한 6G 통신 시스템과 관련된 것이다. 본 개시의 일 실시예에 따르면, 송신단의 커버리지 향상 및 수신단의 수신 성능을 향상시키기 위해, 수신단에서 PA 입출력 관계를 AI 모델에 학습시키고, 학습된 AI 모델을 사용하여 수신된 데이터에 대해 PA 비선형성을 보상하는 방법 및 장치가 제공된다. 또한, 본 개시의 일 실시예에 따르면, AI 모델의 학습이 필요한 경우에만 수신단에 AI 모델을 학습시킬 것을 지시하고, AI 모델의 학습에 적절한 기준 신호를 지시하는 방법 및 장치가 제공된다. 이를 통해, AI 모델의 학습 시 발생하게 되는 시그널링 오버헤드, 연산 오버헤드 또는 전력 소모를 감소시킬 수 있다. 또한, PA 비선형성을 보상할 데이터와 동일 또는 유사한 특성을 갖는 기준 신호가 AI 모델의 학습에 사용됨으로써 AI 모델이 과대적합되는 것을 방지할 수 있다. 따라서, 최적의 PA 비선형성 보상 성능을 갖는 AI 모델을 얻을 수 있다.
Bibliography:Application Number: WO2022KR18867