A Deep Learning Autoencoder for High Throughput and Power Efficient LEO Satellite Systems

Research output: Contribution to conference without publisher/journalPaper without publisher/journalResearchpeer-review

Abstract

Low Earth Orbit satellites are to become one of the biggest suppliers of wireless communication within the next years. With the implementation of 5G and 6G in satellite communication, the need for more power-efficient satellites is needed. This paper exploits recent advances in deep learning to cope with this challenge. The proposed approach is based on an autoencoder using deep neural networks to identify, moderate, and deal with the signal distortion due to the nonlinearity of power amplifiers that are driven into saturation to ensure high power efficiency.
Different from other state-of-the-art results, our architecture can deal with different saturation levels of the front-end PA's while maintaining performance.
Our experimental results for the implemented system demonstrate
a bit error rate comparable to the theoretical bounds of QPSK and M-QAM when using a 100 MHz bandwidth non-linear signal. This exceeds results from comparable systems presented in the literature and suggests the feasibility of high throughput deep learning transceivers that can correct for hardware imperfections through the modulation of the signal.
Original languageEnglish
Publication date30 May 2021
Publication statusSubmitted - 30 May 2021
EventGlobal Communications Conference 2021 - North Convention Center of the IFEMA MADRID Trade Fair of Madrid, Madrid, Spain
Duration: 7 Nov 202111 Dec 2021
https://globecom2021.ieee-globecom.org

Conference

ConferenceGlobal Communications Conference 2021
LocationNorth Convention Center of the IFEMA MADRID Trade Fair of Madrid
CountrySpain
CityMadrid
Period07/11/202111/12/2021
Internet address

Fingerprint

Dive into the research topics of 'A Deep Learning Autoencoder for High Throughput and Power Efficient LEO Satellite Systems'. Together they form a unique fingerprint.

Cite this