LLM4XCE: Large Language Models for Extremely Large-Scale Massive MIMO Channel Estimation

TL;DR

This work addresses the challenge of estimating channels in XL-MIMO under hybrid-field propagation by recasting channel estimation as a semantically-driven task. It introduces LLM4XCE, which fuses pilot features and spatial structure through a Parallel Feature-Spatial Attention module and fine-tunes only the top two Transformer layers of a GPT-2 backbone to produce accurate channel estimates. The approach demonstrates superior NMSE performance and robust generalization across near-field and far-field conditions compared with multiple baselines, highlighting the potential of semantic representations in wireless sensing. The method offers cost-efficient training and a flexible framework for integrating LLMs into high-dimensional MIMO channel estimation.

Abstract

Extremely large-scale massive multiple-input multiple-output (XL-MIMO) is a key enabler for sixth-generation (6G) networks, offering massive spatial degrees of freedom. Despite these advantages, the coexistence of near-field and far-field effects in hybrid-field channels presents significant challenges for accurate estimation, where traditional methods often struggle to generalize effectively. In recent years, large language models (LLMs) have achieved impressive performance on downstream tasks via fine-tuning, aligning with the semantic communication shift toward task-oriented understanding over bit-level accuracy. Motivated by this, we propose Large Language Models for XL-MIMO Channel Estimation (LLM4XCE), a novel channel estimation framework that leverages the semantic modeling capabilities of large language models to recover essential spatial-channel representations for downstream tasks. The model integrates a carefully designed embedding module with Parallel Feature-Spatial Attention, enabling deep fusion of pilot features and spatial structures to construct a semantically rich representation for LLM input. By fine-tuning only the top two Transformer layers, our method effectively captures latent dependencies in the pilot data while ensuring high training efficiency. Extensive simulations demonstrate that LLM4XCE significantly outperforms existing state-of-the-art methods under hybrid-field conditions, achieving superior estimation accuracy and generalization performance.

Figures (6)

• Figure 1: A hybrid-field XL-MIMO system.
• Figure 2: Overview of the proposed LLM4XCE network architecture.
• Figure 3: Illustration of Parallel Feature-Spatial Attention module.
• Figure 4: NMSE versus SNR for the far-field user.
• Figure 5: NMSE versus SNR for the near-field user.