arXiv subjects

Seungmin Choi

Publications and source records attributed to Seungmin Choi.

3 recordsLinked to original sources

arXiv2026

End-to-End Mobility-Aware Multi-RIS Optimization via Blockage Detection and Closed-Form Riemannian Updates

Millimeter-wave (mmWave) multi-user MIMO systems are highly susceptible to dynamic blockages, and reconfigurable intelligent surfaces (RIS) have been introduced as a remedy. However, RIS links can themselves be blocked, while existing studies often assume ideal availability. This paper proposes an end-to-end mobility-aware multi-RIS optimization framework that integrates per-RIS blockage detection with closed-form Riemannian updates. The base station transmits short indexed synchronization signals, enabling each user to identify blocked panels via a simple energy test. Based on the detected feasible sets, we jointly optimize the BS precoder and RIS phases using a Stochastic Closed-form Riemannian Phase Alignment (SCRPA) algorithm, which ensures unit-modulus feasibility, monotone convergence, and low complexity. Extensive simulations validate reliable blockage detection and demonstrate significant weighted sum-rate and scalability gains compared to existing baselines.

eess.SP↗

arXiv2026

Wing-Rotor Aerodynamic Interactions in Small UAVs During Hover and Cruise

A compact vertical take-off and landing aircraft requires the same tilt-rotor configuration to perform two fundamentally different aerodynamic tasks: sustain hover and deliver efficient cruise. This work investigates the underlying wing-rotor interactions in both operating regimes using a validated unsteady Reynolds-averaged Navier-Stokes equations-based computational framework. For cruise, the advance ratio governs the balance between thrust production, propulsive efficiency, and wake coherence. Lower advance ratios produce a tightly wound slipstream that undergoes strong vortex interactions, leapfrogging, and early wake bifurcation. At higher advance ratios, the slipstream retains a narrower and more coherent jet-like structure, improving propulsive efficiency while reducing both thrust of the propellor and lift of the wing. The flow impingement at the wing is characterized through the approaching, interaction, and convection phases, revealing the combined influence of vortex stretching, wake bifurcation, blockage, image-induced velocity, and streamwise momentum convection on the downstream wake. In hover, the propeller's rotational speed governs the overall aerodynamic performance more strongly than the wing's placement. Although the position of the wing with respect to the propellor modifies the local wake interactions and flow impingement on its leading edge, its influence on the integrated thrust coefficient and figure of merit remains limited. We also explain the vortex and wake dynamics around the propellor and the wing responsible for governing these aerodynamic performance.

physics.flu-dyn↗

arXiv2026

Small Models Scout Bottleneck Order for Large-Model Data Control

Small proxy models are commonly used to identify data mixtures for larger-scale training. We ask whether their training trajectories reveal another transferable structure: the order in which larger models should resolve skill bottlenecks. We formulate first-passage skill training, where each monitored skill has a target floor and the objective is to minimize the tokens required to reach all floors. We introduce LogFloor, a closed-loop controller that directs each round toward current bottlenecks, producing phase-ordered resolution trajectories. Across five bAbI skill slices on Qwen2.5-1.5B, LogFloor reduces token cost by 56.2% on average. In 70M-to-12B transfer, three-round replay of a 70M scout path reaches every floor in all eight target runs, saving 30.9% by pair mean, 39.4% in pooled training tokens, and 37.6% under source-cost accounting. On MMLU-control, a frozen scout path succeeds across all eight 12B runs. Collapsing a path to its static marginal mixture or reversing its phase order removes most benefits, while bottleneck labels alone remain partially useful. These results identify phase-ordered bottleneck resolution as a transferable curriculum structure for monitored skill-targeted training.

cs.AI↗