SemantIC: Semantic Interference Cancellation Towards 6G Wireless Communications

Example codes for the paper “SemantIC: Semantic Interference Cancellation Towards 6G Wireless Communications”, which has been accepted for publication in IEEE Communications Letters with DOI: 10.1109/LCOMM.2024.3412973.

arXiv: https://arxiv.org/abs/2310.12768

IEEE Xplore: https://ieeexplore.ieee.org/document/10553320

Instruction

Tested with

• python 3.7.16
• pytorch 1.13.0

Steps

Run with the pre-trained semantic neural network:

• Directly run “SemantIC.py” with the pre-trained semantic neural network “semantic_coder.pkl” to test the semantic interference cancellation systems.

Or training from the beginning:

• Run “googlenet_train.py” to obtain neural network for classifier.
• Run “ENC_DEC_train.py” to obtain neural network for semantic encoder and decoder.
• Run “SemantIC.py” to test the semantic interference cancellation systems.

Notes

The source codes of LDPC are revised from the codes in: https://github.com/hichamjanati/pyldpc

The source codes of example semantic neural network, “googlenet_train.py” and “ENC_DEC_train.py”, are revised from the codes in: https://github.com/SJTU-mxtao/Semantic-Communication-Systems

This framework can be adaptive to other semantic neural network by revising the class “SemanticNN” in “SemantIC.py”.

Citation

BibTeX information:

@Article{lin2024SemantIC,
  author  = {Wensheng Lin and Yuna Yan and Lixin Li and Zhu Han and Tad Matsumoto},
  journal = {IEEE Communications Letters},
  title   = {{SemantIC}: {Semantic} Interference Cancellation Toward {6G} Wireless Communications},
  year    = {2024},
  doi     = {10.1109/LCOMM.2024.3412973}
}