Lightweight Deep Learning-Based Channel Estimation for RIS-Aided Extremely Large-Scale MIMO Systems on Resource-Limited Edge Devices

TL;DR

This work tackles cascaded channel estimation for RIS-aided XL-MIMO in resource-constrained edge settings. It introduces a lightweight deep learning framework that leverages spatial correlation via patch-based training and a multi-scale encoder–decoder with a dedicated denoising module to efficiently recover the cascaded channel $\mathbf{G}_k$ with reduced input dimensionality. The approach demonstrates substantial NMSE gains over traditional LS/LMMSE and prior DL methods across wide SNRs, while maintaining low computational complexity suitable for edge devices, and it scales effectively to large $M$ and $N$. The results on a synthetic but realistic XL-MIMO dataset confirm robustness to spatial correlation, varying patch sizes, and different phase-shift configurations, highlighting practical potential for RIS-enabled 6G and beyond systems.

Abstract

Next-generation wireless technologies such as 6G aim to meet demanding requirements such as ultra-high data rates, low latency, and enhanced connectivity. Extremely Large-Scale MIMO (XL-MIMO) and Reconfigurable Intelligent Surface (RIS) are key enablers, with XL-MIMO boosting spectral and energy efficiency through numerous antennas, and RIS offering dynamic control over the wireless environment via passive reflective elements. However, realizing their full potential depends on accurate Channel State Information (CSI). Recent advances in deep learning have facilitated efficient cascaded channel estimation. However, the scalability and practical deployment of existing estimation models in XL-MIMO systems remain limited. The growing number of antennas and RIS elements introduces a significant barrier to real-time and efficient channel estimation, drastically increasing data volume, escalating computational complexity, requiring advanced hardware, and resulting in substantial energy consumption. To address these challenges, we propose a lightweight deep learning framework for efficient cascaded channel estimation in XL-MIMO systems, designed to minimize computational complexity and make it suitable for deployment on resource-constrained edge devices. Using spatial correlations in the channel, we introduce a patch-based training mechanism that reduces the dimensionality of input to patch-level representations while preserving essential information, allowing scalable training for large-scale systems. Simulation results under diverse conditions demonstrate that our framework significantly improves estimation accuracy and reduces computational complexity, regardless of the increasing number of antennas and RIS elements in XL-MIMO systems.

Figures (9)

• Figure 1: The uplink of the RIS-assisted XL-MIMO system.
• Figure 2: Channel estimation framework has two phases: (1) Direct channel estimation with the RIS off, and (2) Cascaded channel estimation with the RIS on using $L \ge N$ distinct phase shift matrices.
• Figure 3: The proposed framework employs $l$ levels, each with a dedicated denoising block in the encoder. Filters double, and the resolution halves at each level. Fused encoder and decoder features serve as input to each decoder layer.
• Figure 4: The training and inference stages of the proposed framework use matrices of dimension $\mathbf{R}$, where the last dimension represents the real and imaginary parts of the channel.
• Figure 5: Performance comparison of the proposed scheme with benchmark methods and with varying patch sizes.