The paper introduces DRIFT, a lightweight framework for joint channel estimation and prediction in low Earth orbit non-terrestrial networks, aiming to reduce pilot overhead by using data-driven processing after the initial slot. DRIFT refines data-aided channel estimates and forecasts future channel responses with low computational cost, offering two variants based on convolutional and LSTM layers. Simulations show up to 12% spectral efficiency gain over conventional pilot-based systems, with under 200k multiply-accumulate operations suitable for on-board satellite implementation.
By Bruno De Filippo, Carla Amatetti, Alessandro Vanelli-Coralli
The paper proposes a hierarchical Open Radio Access Network (O‑RAN) control framework for tethered millimeter‑wave UAV‑mounted 5G base stations (gNBs). A Non‑Real‑Time RIC application jointly manages UAV placement and slice budgets, while a Near‑Real‑Time RIC application allocates per‑user resources using a permutation‑equivariant DeepSets Soft Actor‑Critic scheduler. This two‑level controller improves eMBB service‑level agreement satisfaction by up to 17 % and URLLC on‑time delivery by up to 42 % compared with conventional schedulers.
By Alireza Mohammadhosseini, Fatemeh Afghah
NS3Learn is a closed‑form model that captures realistic 5G NR sidelink Mode‑2 reception losses—such as half‑duplex loss, scheduling collisions, receiver capture, and decoding—by fitting 10.5 million labeled outcomes from ns‑3 5G‑LENA traces. The model achieves a mean absolute deviation of 0.06 in per‑instant delivery compared to ns‑3, outperforming alternative models, and its parameters transfer with minimal error to new intersections. Using NS3Learn in traffic‑network simulations reverses traffic speed trends and more than doubles predicted hard‑braking events, demonstrating its impact on safety assessments.
By Rasheed Bello, Arthur Mukwaya, Gurcan Comert, Varghese Vaidyan, Vijay Bendigeri, Anthony Dontoh, Jagruti Sahoo, Judith Mwakalonge
CellTwin, also known as NeuralEmu, is a machine‑learning‑based emulation framework that learns 5G scheduler behavior from high‑resolution network telemetry. It predicts resource block allocations and modulation schemes for multiple clients, reconstructs background traffic patterns, and runs as a high‑performance Linux middlebox. Compared to existing tools, it reduces emulation error by up to 57% for key real‑time applications such as web‑page load, WebRTC, and cloud gaming.
By Haoran Wan, Yaxiong Xie, Kyle Jamieson
arXiv:2607. 14975v1 Announce Type: new Abstract: Channel foundation models (CFMs) are developing rapidly, with recent studies reporting benefits from pretraining across downstream wireless tasks.
By Yuan Gao, Wenjun Yu, Jun Jiang, Yunfan Li, Xinyu Guo, Shugong Xu
The paper evaluates eight existing pilot‑aided channel estimators for 5G‑NR and LTE across various 3GPP channel models and numerologies, showing that no single estimator dominates under all conditions. It introduces a condition‑adaptive multi‑agent orchestrator that selects the best estimator per operating scenario, achieving performance within 1.07 dB of an oracle and improving NMSE by up to 3.6 dB at high SNR. The orchestrator runs agents in parallel, yielding near‑single‑estimator latency while scaling wall‑clock time roughly inversely with the number of workers.
By I. Zakir Ahmed, Hamid Sadjadpour
AIR-LLM is an edge inference architecture that broadcasts large language model (LLM) weights over radio, allowing edge devices to perform matrix-vector multiplications directly in the RF domain without storing or loading the weights. The system uses MIMO spatial multiplexing and an energy‑efficient precoder‑postcoder pair to reduce airtime and calibrate the wireless channel, enabling a single broadcast to serve unlimited users. Experiments on real urban channel models show that AIR-LLM achieves only a 4.0% perplexity loss on LLaMA‑3.1‑8B while saving energy by up to 157.7× compared to FP16 and reducing airtime by over 100× for 20 users.
By Zhihui Gao, Tingjun Chen, Dirk Englund
arXiv:2606. 03664v1 Announce Type: cross Abstract: Ultra Reliable and Low Latency Communications (URLLC) was one of the main motivations behind 5G, with 3GPP advertising 1-10 ms latency targets for applications such as industrial automation, Vehicle-To-Everything (V2X), tactical edge networking, and unmanned-system control.
By Maxime Elkael, Michele Polese, Yunseong Lee, Koichiro Furueda, Tommaso Melodia
arXiv:2609.30629v1 Announce Type: cross
Abstract: Mission-critical UAVs increasingly rely on split vision-language model (VLM) perception under tight onboard-resource and wireless-communication const...
By Rajat Bhattacharjya, Minwoo Kim, Arnab Sarkar, Tamoghno Das, Sing-Yao Wu, Eli Bozorgzadeh, Marco Levorato, Nikil Dutt
arXiv:2602. 06932v5 Announce Type: replace Abstract: Speculative decoding can significantly accelerate LLM serving, yet most deployments today disentangle speculator training from serving, treating speculator training as a standalone offline modeling problem.
By Junxiong Wang, Fengxiang Bie, Jisen Li, Zhongzhu Zhou, Zelei Shao, Yubo Wang, Yinghui Liu, Qingyang Wu, Avner May, Sri Yanamandra, Ce Zhang, Tri Dao, Percy Liang, Ben Athiwaratkun, Shuaiwen Leon Song, Chenfeng Xu, Xiaoxia Wu
arXiv:2603. 18853v3 Announce Type: replace-cross Abstract: Autonomous aerial vehicles (AAVs) enable data collection for sixth-generation Internet-of-Things networks, but their trajectories couple nonlinear wireless rates with long-horizon service progress.
By Xiucheng Wang, Zhenye Chen, Nan Cheng, Zhisheng Yin, Xuemin Shen
The paper introduces VLA-ULAP, a lightweight local action predictor that interleaves remote vision–language–action (VLA) calls with on‑edge inference. ULAP, with only 7.4 M parameters, predicts action chunks in a single pass using current views, proprioception, and action history, eliminating the need for VLA hidden states or server round‑trips. Experiments on Jetson Orin Nano and simulated benchmarks show that VLA-ULAP can remove 48.8–76.7 % of VLA calls while preserving 95–97.5 % of baseline success, and it outperforms local VLA‑acceleration alternatives in both inference time and energy consumption.
By Deyu Cao, Ryuji Oi, Kosuke Matsushima, Yuxuan Pan, Ziheng Wang, Daichi Fujiki, Atsutake Kosuge