Low Complexity Residual Learning for Pilot Signal Reconstruction in RIS-Aided MIMO Systems
IEEE Communications Letters · 2026 · 30 · 2367-2371
DOI: 10.1109/lcomm.2026.3707385 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Communications Letters, 2026. Verfasst von Lazaros A. Iliadis und 6 weiteren Autor:innen. Die Arbeit nennt 16 Literaturangaben.
Abstract (Originalsprache)
This work proposes an efficient, low complexity structure-aware residual neural network (ResNet) for pilot signal reconstruction in reconfigurable intelligent surface (RIS)-aided multiple-input multiple-output (MIMO) systems. The proposed approach operates directly on the received pilot tensor and performs signal-domain denoising, avoiding explicit estimation of the individual base station (BS)–RIS and RIS–user equipment (UE) channel components. By leveraging the low-dimensional structure induced by sparse millimeter wave (mmWave) propagation and RIS phase coupling, the framework accurately reconstructs the cascaded RIS-aided MIMO response from noisy pilot observations. To enable hardware-efficient deployment, the ResNet is further compressed via structured channel pruning and post-training INT8 quantization. Numerical results demonstrate that the resulting model, trained in a single signal-to-noise ratio (SNR) of 20 dB, generalizes effectively across a wide range of unseen SNR conditions without retraining, outperforming both classical model-based approaches and representative data-driven baselines. The proposed quantized model is evaluated under realistic computational constraints, making it suitable for real-time implementation in future wireless networks.
Zitieren
APA
Iliadis, L. A., Zaharis, Z. D., Sarigiannidis, P., Karagiannidis, G. K., Christodoulou, C. G., Renzo, M. D., & Goudos, S. K. (2026). Low Complexity Residual Learning for Pilot Signal Reconstruction in RIS-Aided MIMO Systems. IEEE Communications Letters, 30, 2367-2371. https://doi.org/10.1109/lcomm.2026.3707385
Vancouver
Iliadis LA, Zaharis ZD, Sarigiannidis P, Karagiannidis GK, Christodoulou CG, Renzo MD, et al. Low Complexity Residual Learning for Pilot Signal Reconstruction in RIS-Aided MIMO Systems. IEEE Communications Letters. 2026;30:2367-2371. doi:10.1109/lcomm.2026.3707385
BibTeX
@article{Iliadis2026,
author = {Iliadis, Lazaros A. and Zaharis, Zaharias D. and Sarigiannidis, Panagiotis and Karagiannidis, George K. and Christodoulou, Christos G. and Renzo, Marco Di and Goudos, Sotirios K.},
title = {Low Complexity Residual Learning for Pilot Signal Reconstruction in RIS-Aided MIMO Systems},
journal = {IEEE Communications Letters},
year = {2026},
volume = {30},
pages = {2367--2371},
doi = {10.1109/lcomm.2026.3707385}
}
EUVIMED