AI 中文总结
针对MIMO系统，提出基于大语言模型的语义感知数据辅助信道估计框架，利用语义信息进行可靠符号选择与校正，采用两层机制，仿真表明该框架在归一化均方误差和误码率上优于传统方案，接近理想估计器性能。
AI 中文摘要
数据辅助信道估计通过将检测到的符号用作虚拟导频来提高频谱效率。在此过程中，仅选择可靠符号对于防止误检测符号破坏信道估计至关重要。然而，传统方法仅依赖物理层统计信息。除物理层信息外，传输的有效载荷具有可用于解决检测错误的固有语义结构。本文提出了一种用于多输入多输出（MIMO）系统的新型语义感知信道估计框架，该框架利用微调的大语言模型（LLM）基于语义信息进行可靠符号选择和校正。该框架采用两层机制：一层通过语义验证选择可靠的解码符号，另一层通过使用物理层信息将其与接收信号进行交叉验证来选择经LLM准确校正的符号。我们证明，校正后的符号在估计误差方面产生的预期减少量比最初正确解码的符号严格更大。大量仿真表明，所提出的框架在归一化均方误差和误码率方面均明显优于传统的数据辅助方案，性能接近理想估计器。
英文摘要
Data-aided channel estimation enhances spectral efficiency by reusing detected symbols as virtual pilots. In this process, selecting only reliable symbols is crucial to prevent misdetected symbols from corrupting the channel estimate. However, conventional methods rely exclusively on physical-layer statistics. Beyond physical-layer information, transmitted payloads possess inherent semantic structures that can be exploited to resolve detection errors. In this paper, we propose a novel semantic-aware channel estimation framework for multiple-input multiple-output (MIMO) systems that utilizes a fine-tuned large language model (LLM) to perform reliable symbol selection and correction based on semantic information. The framework employs a two-layer mechanism: one layer selects reliable decoded symbols through semantic verification, while the other selects accurately LLM-corrected symbols by cross-validating them against the received signal using physical-layer information. We prove that corrected symbols yield a strictly larger expected reduction in estimation error than initially correctly decoded symbols. Extensive simulations demonstrate that the proposed framework significantly outperforms conventional data-aided schemes in both normalized mean squared error and bit error rate, closely approaching the performance of an oracle estimator.
