arXiv Science⌕ Search
Home/Archive/arXiv

arXiv · 2610.15404

Eunoia: Toward Application-Transparent, Scalable GPU Sharing in AI-RAN

Zongshen Wu·Marko Milenković·Jiarong Xing·Rahman Doost-Mohammady
View original source ↗Read source document ↗Export metadata ↓

Abstract

The AI-RAN paradigm makes the next-generation cellular network more versatile by co-locating AI and RAN workloads on the same commodity hardware. While the GPU is a promising platform for AI-RAN realization, a generic concurrent AI application interferes with the latency-sensitive RAN on GPU compute resources. In this paper, we propose Eunoia, a dynamic real-time AI-RAN GPU scheduler that is transparent to both AI applications and the RAN workload and operates at the granularity level of the RAN slot. We deploy Eunoia onto NVIDIA Aerial AI-RAN and evaluate Eunoia on both an over-the-air 5G private network and an emulated multi-cell network. We demonstrate that in a 27-cell AI-RAN, Eunoia keeps the RAN uplink P99 latency under 3 ms while delivering higher AI performance than other RAN-safe static partitioning methods.

Explore related subjects

eess.SP

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Zongshen Wu, Marko Milenković, Jiarong Xing, Rahman Doost-Mohammady. 2026-10-11. Eunoia: Toward Application-Transparent, Scalable GPU Sharing in AI-RAN. https://arxiv.org/abs/2610.15404

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related papers

arXiv2026

Convexity Meets Curvature: Lifted Near-Field Super-Resolution

Extra-large apertures, high carrier frequencies, and integrated sensing and communications (ISAC) are pushing array processing into the Fresnel region, where spherical wavefronts induce a range-dependent phase across the aperture. This curvature breaks the Fourier/Vandermonde structure behind classical subspace methods, and it is especially limiting with hybrid front-ends that provide only a small number of pilot measurements. Consequently, practical systems need continuous angle resolution and joint angle-range inference where many near-field approaches still rely on costly 2D gridding. We show that convexity can meet curvature via a lifted, gridless superresolution framework for near-field measurements. The key is a Bessel-Vandermonde factorization of the Fresnel-phase manifold that exposes a hidden Vandermonde structure in angle while isolating the range dependence into a compact coefficient map. Building on this, we introduce a lifting that maps each range bin and continuous angle to a structured rank-one atom, converting the nonlinear near-field model into a linear inverse problem over a row-sparse matrix. Recovery is posed as atomic-norm minimization and an explicit dual characterization via bounded trigonometric polynomials yields certificate-based localization that super-resolves off-grid angles and identifies active range bins. Simulations with strongly undersampled hybrid observations validate reliable joint angle-range recovery for next-generation wireless and ISAC systems.

eess.SP↗
arXiv2026

Joint and Streamwise Distributed MIMO Satellite Communications with Multi-Antenna Ground Users

We consider a low Earth orbit downlink communication, where multiple satellites jointly serve multi-antenna ground users, transmitting multiple spatial streams per user. Using a line-of-sight-dominant satellite channel model with statistical channel state information, including angular information and large-scale fading, we study two distributed transmission modes with different fronthaul requirements. First, for joint transmission, where all satellites transmit all user streams, we formulate a sum spectral efficiency (SE) maximization problem under general convex power constraints and address the intractability of the exact ergodic SE expression by adopting a tractable approximation. Exploiting the equivalence between sum SE maximization and weighted sum mean square error minimization, we derive a novel iterative transceiver design. Second, to reduce fronthaul load, we propose streamwise transmission, where each stream is sent by a single satellite, and develop an eigenmode-based stream-satellite association using participation factors and a maximum-weight bipartite matching problem solved by the Hungarian algorithm. Numerical simulations evaluate the validity of the SE approximation, demonstrate conditions under which streamwise transmission performs nearly optimally or trades SE for lower overhead, highlight the impact of stream/user loading, and show substantial performance gains over conventional benchmarks.

eess.SP↗
arXiv2026

A Ray-Based Characterization of Satellite-to-Urban Propagation

The evolution toward 6G communication systems is expected to rely on integrated three-dimensional network architectures where terrestrial infrastructures coexist with non-terrestrial stations such as satellites, enabling ubiquitous connectivity and service continuity. In this context, accurate channel models for satellite-to-ground propagation in urban environments are essential, particularly for user equipment located at street level where obstruction and multipath effects are significant. This work investigates satellite-to-urban propagation through deterministic ray-tracing simulations. Three representative urban layouts are considered, namely dense urban, urban, and suburban. Multiple use cases are investigated, including handheld devices, vehicular terminals, and fixed rooftop receivers operating across several frequency bands. The analysis focuses on the relative importance of competing propagation mechanisms and on four key channel parameters, namely shadowing standard deviation, Rician \(K\)-factor, delay spread and Doppler spread, which are relevant for the calibration of channel models to be used in link- and system-level simulations. Results show that elevation and receiver placement strongly shape fading and dispersion, antenna directivity affects shadowing differently under LoS and NLoS conditions, urban morphology produces nontrivial multipath trends, and terminal mobility largely governs Doppler spread.

eess.SP↗