Year: 2026
Type: conference-paper
Abstract: Reconfigurable intelligent surfaces (RIS) are emerging as a key enabler for mmWave systems by providing programmable reflections that enhance coverage and shape propagation. Accurate localization is essential for beam alignment, mobility management, and integrated sensing--communication, yet it is challenging at mmWave due to high directionality and sparse multipath. In this work, we exploit RIS beam control to facilitate localization by generating distinctive SNR fingerprints across a small set of beam-sweeping configurations. Based on a base station (BS)--RIS--user equipment (UE) geometry, we construct a supervised learning framework that maps the measured signal to noise ratio (SNR) responses to the user polar coordinates set in a fixed azimuth plane, without introducing additional physical equations beyond the adopted channel model. The proposed approach achieves centimeter-level ranging performance, reaching a minimum mean absolute error (MAE) of <4 cm at 28 GHz with a BS antenna array of 20 X 20 antenna elements, while maintaining strong SNR prediction fidelity over the considered angular--range variations. These results highlight the potential of RIS-assisted SNR fingerprinting as a low-overhead and practical solution for future mmWave networks, enabling reliable localization to support beamforming, tracking, and RIS-enabled integrated sensing and communication (ISAC) functionalities.
Institution: Queen's University Belfast
Language: English
Subfield: Electrical and Electronic Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: green
Awards: G1213672793, G2317614305, G4348145101
OpenAlex ID: W7150388686
FWCI:
Cites:
Cited by:
Related to: