RIS-Assisted Beamfocusing in Near-Field IoT Communication Systems: A Transformer-Based Approach

TL;DR

This work tackles near-field IoT communications with ELAA by deploying RIS-assisted beamfocusing and a Transformer-based two-stage beam training approach that jointly explores angular and distance domains. By converting beam code selection into a device-position mapping problem, the method achieves precise beamfocusing and reduces training overhead. The proposed RIS-enabled architecture and MSE-driven transformer training yield beam selection accuracies up to $0.97$ at $20$ dB and offer $10\%$ to $50\%$ gains over baselines across SNRs. This approach demonstrates the practical potential of RIS-enabled NFC for scalable, high-capacity IoT in future wireless networks.

Abstract

The massive number of antennas in extremely large aperture array (ELAA) systems shifts the propagation regime of signals in internet of things (IoT) communication systems towards near-field spherical wave propagation. We propose a reconfigurable intelligent surfaces (RIS)-assisted beamfocusing mechanism, where the design of the two-dimensional beam codebook that contains both the angular and distance domains is challenging. To address this issue, we introduce a novel Transformer-based two-stage beam training algorithm, which includes the coarse and fine search phases. The proposed mechanism provides a fine-grained codebook with enhanced spatial resolution, enabling precise beamfocusing. Specifically, in the first stage, the beam training is performed to estimate the approximate location of the device by using a simple codebook, determining whether it is within the beamfocusing range (BFR) or the none-beamfocusing range (NBFR). In the second stage, by using a more precise codebook, a fine-grained beam search strategy is conducted. Experimental results unveil that the precision of the RIS-assisted beamfocusing is greatly improved. The proposed method achieves beam selection accuracy up to 97% at signal-to-noise ratio (SNR) of 20 dB, and improves 10% to 50% over the baseline method at different SNRs.

Figures (12)

• Figure 1: Illustration of RIS-assisted near-field beamfocusing between the BS and device.
• Figure 2: The proposed two-stage codebook for near-field beam training.
• Figure 3: The proposed Transformer-based architecture for near-field beam training.
• Figure 4: The variation curves of loss values on the validation set and training set with increasing number of trainings under different SNRs.
• Figure 5: The variation curves of accuracy values on the validation set with increasing number of training epochs under different SNRs.