Deep Learning for Advanced Physical Layer Communications

Driven by the booming amount of transmission information, modern communication systems have developed to the fifth generation, and are expected to integrate different radio access technologies, including the satellite component. As outlined in the 3GPP, the future integrated satellite and terrestrial architectures will lead to manifold advantages and make satellite communications essential to the evolution fo the 5G network. Thus, the future 5G satellite network is required to have low latency, high capacity, and strong adaptability to complex environments, and these requirements are aligned to our research objectives, “developing advanced communication networks which are efficient and stable”.

However, the traditional satellite communication systems only have a limited ability to face challenges in 5G scenarios, including high attenuation, the complex and unreliable communication environment and resulting transmission errors. Though there are some existing solutions for satellite communications to combat these channel impairments, they cannot work well enough for 5G. To ensure the accuracy, efficiency and reliability of the future wireless communications system, our research aims to develop DL based wireless physical layer frameworks (i.e., leveraging deep-learning to redesign the module of the conventional communication system) for performance improvement, which can also be used to implement 5G satellite communications.


Project Leader:
Professor Yonghui Li, University of Sydney

PhD Student:
Kou Tian, University of Sydney