The symbol-level precoding (SLP) technique can enhance the performance in multi-user wireless communication systems because of its ability to convert harmful multi-user interference (MUI) into beneficial ones. However, the tremendous computational complexity of conventional symbol-level precoding designs severely hinders practical implementations. This paper proposes an SLP design scheme based on unsupervised learning in a multiple-input multiple-output (MIMO) downlink system. In the SLP design scheme, the loss function is first designed to improve performance by pushing the received signal further away from the decision boundaries into a constructive region. An efficient symbol-level precoding network (SLP-Net) is introduced to optimize the SLP under the power constraint and adapt amplitude phase modulation. Numerical results highlight that the optimized SLP design scheme provides meaningful performance gain in the MIMO downlink system. The proposed SLP design scheme can be an efficient technology to perform better in the future 6G.
Unsupervised Learning Based Symbol-Level Precoding Design for Amplitude Phase Modulation
07.10.2024
787985 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
SYMBOL VS BLOCK LEVEL PRECODING IN MULTI-BEAM SATELLITE SYSTEMS
TIBKAT | 2019
|Phase precoding with constant amplitude for slow fading channels
Kraftfahrwesen | 1995
|