Accurate channel prediction is essential in massive multiple-input multiple-output (m-MIMO) systems to improve precoding effectiveness and reduce the overhead of channel state information (CSI) feedback. However, existing methods often suffer from accumulated prediction errors and poor generalization to dynamic wireless environments, making it challenging to maintain high prediction accuracy. Large language models (LLMs) have demonstrated remarkable modeling and generalization capabilities in tasks such as time series prediction, making them a promising solution. Nevertheless, a significant modality gap exists between the linguistic knowledge embedded in pretrained LLMs and the intrinsic characteristics of CSI, posing substantial challenges for their direct application to channel prediction. Moreover, the large parameter size of LLMs hinders their practical deployment in real-world communication systems with stringent latency constraints. To address these challenges, we propose a novel channel prediction framework based on semantically aligned large models, referred to as CSI-ALM, which bridges the modality gap between natural language and channel information. Specifically, we design a cross-modal fusion module that aligns CSI representations with the language feature space using a pretrained corpus. Additionally, we maximize the cosine similarity between word embeddings and CSI embeddings to construct semantic cues, effectively leveraging the latent knowledge in LLMs. To reduce complexity and enable practical implementation, we further introduce a lightweight version of the proposed approach, called CSI-ALM-Light. This variant is derived via a knowledge distillation strategy based on attention matrices, which extracts essential features from the teacher model, CSI-ALM, and transfers them to a compact, efficient student model, CSI-ALM-Light. Extensive experimental results demonstrate that CSI-ALM consistently outperforms state-of-the-art deep learning methods across various communication scenarios, achieving substantial performance gains. Moreover, under limited training data conditions—where all models are trained using only 10% of the original training dataset—CSI-ALM-Light, with only 0.34M parameters, attains performance comparable to CSI-ALM and significantly outperforms conventional deep learning approaches. These validate the effectiveness of the proposed approach for accurate and efficient channel prediction in m-MIMO systems.

Large language models (LLMs) deliver strong performance across diverse tasks but their growing computational demands make single-device inference increasingly infeasible. Distributed inference on heterogeneous accelerators offers a scalable solution, yet existing partitioning strategies often depend on oversimplified latency models that fail to capture runtime variability caused by device diversity and fluctuating input lengths. We present an adaptive framework that jointly predicts submodel latency and optimally partitions LLMs to minimize end-to-end inference time in heterogeneous clusters. A multilayer perceptron (MLP) regression model integrates multidimensional features—such as submodel masks, device characteristics, token counts, and computational demand—and employs lightweight adapter layers for rapid fine-tuning to unseen environments. Using these latency predictions, we formulate pipeline-aware model partitioning as a dynamic programming problem, reducing search complexity from exponential to polynomial. Experiments on GPT2 show that our approach achieves high prediction accuracy (91.8% within 20% error) and improves adaptability, raising accuracy from 20.3% to 76.0% with only 50 fine-tuning samples. Overall, the proposed method reduces inference latency from 234.70 ms to 130.45 ms, significantly outperforming baseline loadbalancing methods.

Timely transmission of remote sensing images over satellite channels is crucial for applications like real-time monitoring and remote rescue. However, challenges such as limited bandwidth, rain attenuation, long delays, and co-channel interference impede efficient transmission. To address these, we propose SatSemCom, a novel channel-adaptive semantic communication method for remote sensing image transmission that integrates OFDM modulation. SatSemCom features a Vision Transformer (ViT)-based semantic encoder and decoder for compact semantic extraction, along with a channel predictor to predict satellite channel states and mitigate the impact of outdated channel state information. Additionally, a rate adaptation module dynamically adjusts the transmission rate based on channel conditions. Simulation results show that the proposed scheme is effective and robust under fast-changing satellite channels. Compared to existing channel-adaptive SemCom methods, our approach achieves a PSNR improvement of over 1.5 dB in the reconstructed high-resolution remote sensing images.