Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

SemantIC: Semantic Interference Cancellation Towards 6G Wireless Communications

Wensheng Lin, Yuna Yan, Lixin Li, Zhu Han, Tad Matsumoto

TL;DR

SemantIC presents a receiver-centric method to cancel semantic and signal-domain interference by attaching a semantic auto-encoder to a conventional channel decoder, forming a turbo loop guided by Wyner-Ziv-inspired side information stored in neural network parameters. The approach denoises in the semantic domain and re-injects improved priors into iterative LDPC decoding, achieving higher information quality without consuming additional channel resources. Experiments on CIFAR-10 image transmission show consistent BER and PSNR gains, particularly in low-SNR regimes, with rapid convergence around four iterations. The work highlights a promising direction for 6G where semantic information and side-information-aware decoding can enhance reliability, while noting avenues for further performance boosts via advanced neural modules.

Abstract

This letter proposes a novel anti-interference technique, semantic interference cancellation (SemantIC), for enhancing information quality towards the sixth-generation (6G) wireless networks. SemantIC only requires the receiver to concatenate the channel decoder with a semantic auto-encoder. This constructs a turbo loop which iteratively and alternately eliminates noise in the signal domain and the semantic domain. From the viewpoint of network information theory, the neural network of the semantic auto-encoder stores side information by training, and provides side information in iterative decoding, as an implementation of the Wyner-Ziv theorem. Simulation results verify the performance improvement by SemantIC without extra channel resource cost.

SemantIC: Semantic Interference Cancellation Towards 6G Wireless Communications

TL;DR

SemantIC presents a receiver-centric method to cancel semantic and signal-domain interference by attaching a semantic auto-encoder to a conventional channel decoder, forming a turbo loop guided by Wyner-Ziv-inspired side information stored in neural network parameters. The approach denoises in the semantic domain and re-injects improved priors into iterative LDPC decoding, achieving higher information quality without consuming additional channel resources. Experiments on CIFAR-10 image transmission show consistent BER and PSNR gains, particularly in low-SNR regimes, with rapid convergence around four iterations. The work highlights a promising direction for 6G where semantic information and side-information-aware decoding can enhance reliability, while noting avenues for further performance boosts via advanced neural modules.

Abstract

This letter proposes a novel anti-interference technique, semantic interference cancellation (SemantIC), for enhancing information quality towards the sixth-generation (6G) wireless networks. SemantIC only requires the receiver to concatenate the channel decoder with a semantic auto-encoder. This constructs a turbo loop which iteratively and alternately eliminates noise in the signal domain and the semantic domain. From the viewpoint of network information theory, the neural network of the semantic auto-encoder stores side information by training, and provides side information in iterative decoding, as an implementation of the Wyner-Ziv theorem. Simulation results verify the performance improvement by SemantIC without extra channel resource cost.
Paper Structure (12 sections, 4 equations, 6 figures)

This paper contains 12 sections, 4 equations, 6 figures.

Table of Contents

  1. Introduction
  2. Semantic Interference Cancellation Principle
  3. Theoretical Basis
  4. Decoder Structure
  5. Auto-Encoder
  6. Turbo Loop
  7. Decoder Implementation Design
  8. Simulation Results
  9. Anti-Interference Performance
  10. Iterative Gain
  11. Visual Effect
  12. Conclusion

Figures (6)

  • Figure 1: The system model from the viewpoint of network information theory. $X^n$, $Y^n$ and $\hat{X}^n$ stand for the source information, the side information and the reconstructed information, respectively, with the sequence length of $n$. $M$ is the codeword satisfying the coding rate $R$.
  • Figure 2: The structure of SemantIC system.
  • Figure 3: A simple structure of semantic encoder/decoder.
  • Figure 4: The impact of SNR on BER, , and PSNR.
  • Figure 5: Convergence behaviour for BER, , and PSNR.
  • ...and 1 more figures