Experimental RIS localisation paper published in IEEE Xplore
Our paper “Experimental Validation of Localisation in RIS-Assisted mmWave Networks,” presented at UCMMT 2026 in Birmingham, is now published in IEEE Xplore. It evaluates whether machine-learning localisation models trained on simulated data can transfer to measurements from a physical 28 GHz RIS-assisted system without retraining.
Paper results
- Used a 20 × 20 RIS at 28 GHz with four angularly steered, 1-bit phase profiles.
- KNN achieved 0.94° angle MAE and 5 cm range MAE on the simulation test set.
- Experimental validation achieved 0.5° angle error and 5.3–8.3 cm range error without retraining.
Research experience
Publishing this experimental work brings together the simulation, measurement, and conference presentation stages of the study. The results show how a localisation model trained in simulation performs on measured near-field RIS data.
Next work
The next stage is to test the approach across more user positions and changing propagation conditions to assess its robustness in practical deployments.
Publication citation
M. T. Hassan, D. Zelenchuk, G. Travers, M. A. B. Abbasi, and G. G. Machado, “Experimental Validation of Localisation in RIS-Assisted mmWave Networks,” in Proc. 19th IEEE United Conference on Millimetre Waves and Terahertz Technologies (UCMMT), Birmingham, U.K., 2026.
