Environment-Aware MIMO Channel Estimation in Pilot-Constrained Upper Mid-Band Systems
Seyed Alireza Javid, Nuria González-Prelcic
TL;DR
This work tackles the challenge of accurate MIMO channel estimation under pilot constraints in upper mid-band systems by introducing a physics-informed neural network (PINN) that fuses a coarse LS channel estimate with RSS maps derived from Maxwell-based propagation. The architecture combines a cross-attention enhanced U-Net with Transformer modules to integrate environmental propagation information, yielding substantial NMSE improvements (over 5 dB in some regimes) and robustness across frequencies and environments with limited fine-tuning. Key contributions include the formal linking of physical and channel models via RSS and per-tap field sums, a practical three-stage estimation pipeline, and transfer-learning demonstrations that adapt across bands and urban scenarios. The results suggest that physics-informed priors can significantly improve pilot-efficient channel estimation in massive MIMO, enabling real-time, interpretable deployments in urban upper mid-band networks, with a dataset released to accelerate further research.
Abstract
Accurate multiple-input multiple-output (MIMO) channel estimation is critical for next-generation wireless systems, enabling enhanced communication and sensing performance. Traditional model-based channel estimation methods suffer, however, from performance degradation in complex environments with a limited number of pilots, while purely data-driven approaches lack physical interpretability, require extensive data collection, and are usually site-specific. This paper presents a novel physics-informed neural network (PINN) framework that combines model-based channel estimation with a deep network to exploit prior information about the propagation environment and achieve superior performance under pilot-constrained scenarios. The proposed approach employs an enhanced U-Net architecture with cross-attention mechanisms to fuse initial channel estimates with received signal strength (RSS) maps to provide refined channel estimates. Comprehensive evaluation using realistic ray-tracing data from urban environments demonstrates significant performance improvements, achieving over 5 dB gain in normalized mean squared error (NMSE) compared to state-of-the-art methods, with particularly strong performance in pilot-limited scenarios and robustness across different frequencies and environments with only minimal fine-tuning. The proposed framework maintains practical computational complexity, making it viable for massive MIMO systems in upper mid-band frequencies.