Autoencoder-based end-to-end learning optimizes geometric constellation shapes and bit mappings, achieving up to 0.2 bits per QAM symbol GMI gain across data rates under transceiver impairments.
We show that the proposed autoencoder arrives at a Gray-like code, which does not exhibit this problem
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.IT 1years
2019 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
End-to-end Learning for GMI Optimized Geometric Constellation Shape
Autoencoder-based end-to-end learning optimizes geometric constellation shapes and bit mappings, achieving up to 0.2 bits per QAM symbol GMI gain across data rates under transceiver impairments.