Semantic Pilot Design for Data-Aided Channel Estimation Using a Large Language Model
Sojeong Park, Hyun Jong Yang
TL;DR
This work introduces a semantic pilot design for data-aided channel estimation in text-enabled transmissions by leveraging an LLM to correct decoded text and identify reliably decoded symbols. The semantic pilot, derived from symbols corresponding to unchanged characters after correction, is used with a two-step refinement—phase refinement via LS and magnitude scaling via a scaling factor $\gamma$—to produce a refined channel estimate $\hat{h}_{LLM}$. Simulations on a SISO link with a Rician channel show that the proposed approach yields lower NMSE, smaller phase error, and reduced BER compared to pilot-only and other data-aided methods, with high semantic-pilot reliability and substantial payload usage as pilots. The method is modular and can complement existing data-aided schemes, with future work extending to MIMO and real-time deployment using lightweight LLMs for semantic guidance in channel estimation.
Abstract
This paper proposes a semantic pilot design for data-aided channel estimation in text-inclusive data transmission, using a large language model (LLM). In this scenario, channel impairments often appear as typographical errors in the decoded text, which can be corrected using an LLM. The proposed method compares the initially decoded text with the LLM-corrected version to identify reliable decoded symbols. A set of selected symbols, referred to as a semantic pilot, is used as an additional pilot for data-aided channel estimation. To the best of our knowledge, this work is the first to leverage semantic information for reliable symbol selection. Simulation results demonstrate that the proposed scheme outperforms conventional pilot-only estimation, achieving lower normalized mean squared error and phase error of the estimated channel, as well as reduced bit error rate.