A Lightweight Bi-LSTM-Attention Framework for Efficient Semantic Communication in 6G Edge Networks

The rapid development of 6G networks has established Semantic Communication (SemCom) as a transformative paradigm for high-efficiency data transmission. While recent advances, particularly those leveraging large language models (LLMs) as discussed in recent literature, provide high semantic accuracy, they impose significant computational overhead, making them unsuitable for resource-constrained edge devices. In this paper, we propose a lightweight alternative based on bidirectional long short-term memory (BI-LSTM) networks integrated with an attention mechanism. We evaluate our model on the Ag News dataset, focusing on comparative analysis with LLM-enabled semantic systems. Experimental results show that our proposed model achieves a competitive accuracy of 91.54% with an inference latency of only 1.78 ms, which significantly outperforms the LLM-based baseline in terms of computational efficiency. Furthermore, we assess the robustness of our architecture under different levels of semantic noise (word masking). The findings show that our lightweight Bi-LSTM approach provides an optimal trade-off between performance and resource consumption, providing a practical solution for AI-native communication in low-power environments.