The paper tackles the challenge of converting uplink to downlink channel covariance matrices in FDD massive MIMO systems, where learning‑based methods lose accuracy as the antenna count grows. It identifies that the mean angle of arrival creates a phase ramp whose oscillation rate increases with array size, making fixed‑size datasets sparse. The authors propose a deramping technique that estimates and removes this ramp before learning, reducing estimation error across three different learners and maintaining superiority over model‑based benchmarks at large array sizes.
By Melih Can Zerin
The paper introduces Tri-PNet, a deep learning framework that jointly designs electromagnetic (EM), analog, and digital precoding for tri-hybrid multi-user MIMO OFDM systems. By leveraging Conformer architecture, Tri-PNet learns to select radiation patterns and perform hybrid precoding, achieving higher spectral efficiency than random EM precoding and conventional hybrid MIMO while maintaining low online complexity. The study evaluates two radiation-pattern modes—non-regular and 3GPP TR 38.901—showing robust performance even with imperfect channel state information.
By Kaijun Feng, Jiaxin He, Hongrui Yu, Zhen Gao, Anwen Liao, Ziwei Wan, Zhaocheng Wang
arXiv:2607. 00860v1 Announce Type: cross Abstract: Millimeter-wave (mmWave) beam alignment plays a critical role in next-generation wireless systems, yet its efficient implementation remains challenging.
By Ahmet Nuri Cevik, Sinem Coleri
arXiv:2609.32923v2 Announce Type: replace
Abstract: Wireless digital twins (DTs) rely on 3D environment models to predict radio propagation and support wireless-network decisions, yet these models ar...
By Saad Masrur, Saeed R. Khosravirad, Ismail Guvenc
arXiv:2608. 14676v1 Announce Type: cross Abstract: In 5G/6G wireless systems, accurate and timely channel estimation is critical to ensure reliable communication under complex, fast-changing radio conditions.
By Javad Zolfaghari-Bengar, Rakibul Rony, Elisa Gomez-de-Lope, Alejandro Villena-Rodriguez, Abhinav Mahadevan, Nicolas Kourtellis
arXiv:2608. 14709v1 Announce Type: cross Abstract: This demo presents real-time AI-based uplink channel-estimation inference using data collected from a hardware-in-the-loop 5G platform.
By Javad Zolfaghari-Bengar, Rakibul Rony, Elisa Gomez-de-Lope, Alejandro Villena-Rodriguez, Abhinav Mahadevan, Nicolas Kourtellis
arXiv:2605.00020v2 Announce Type: replace-cross
Abstract: The success of large foundation models is catalyzing a new paradigm for AI-native 6G network design: wireless foundation models for physical-...
By Kejia Bian, Meixia Tao, Jianhua Mo, Zhiyong Chen, Leyan Chen
arXiv:2605.00968v2 Announce Type: replace-cross
Abstract: Wireless foundation models (WFMs) have emerged as a promising paradigm for unified channel state information (CSI) acquisition across diverse...
By Chenyu Zhang, Xinchen Lyu, Chenshan Ren, Yanzhao Hou, Xuefei Zhang, Shuhan Liu, Qimei Cui
arXiv:2609.22139v1 Announce Type: cross
Abstract: Automatic modulation classification (AMC) of received radio signals is prudent for further signal processing tasks such as communication monitoring,...
By Qamar Ijaz, Nayyer Aafaq
arXiv:2610.19015v1 Announce Type: new
Abstract: Real measurements are too limited to train an accurate indoor pathloss predictor that works across a wide range of buildings. Large-scale synthetic pre...
By Khoren Petrosyan, Artashes Mkrtchyan, Rafayel Mkrtchyan, Hrant Khachatrian, Theofanis P. Raptis
arXiv:2608.23469v1 Announce Type: cross
Abstract: A machine learning-assisted framework for the inverse design of pixelated millimetre-wave patch antennas targeting the 22--30 GHz band is presented....
By Nadeem Rather, Holger Claussen, Lester Ho
The paper introduces FuseAMC, a lightweight decision‑fusion model that combines a residual CNN and a gradient‑boosted‑tree cumulant classifier for automatic modulation classification. By training a logistic‑regression meta‑combiner on the class‑posteriors of both models, FuseAMC achieves higher overall accuracy than either approach alone, especially improving performance at both low and high signal‑to‑noise ratios. The resulting system contains only about 0.2 million parameters, occupies a few megabytes, and requires roughly 3 MFLOPs per frame, enabling training and inference on a single CPU core suitable for embedded processors.
By Zhiyi Yin, Zhizhao Wen, Ruoxin Zhang, Chao Wang, Puyang Xu, Yifan Jiang, Yusi Gong