Kawon Han

Pouya Parsa, Kawon Han, Seongjin Choi

Frontier autonomous-driving models -- especially vision-language-action (VLA) models, whose forward pass approaches $\sim$60~TFLOPs -- are outgrowing economical onboard deployment, since peak hardware sits idle most of the day. Cloud inference can instead share GPUs across active vehicles, but the vehicle must upload through a capacity-limited uplink, reach a GPU without queueing, and return a decision within the closed-loop budget. This paper asks: can the cloud drive? We answer with an analytical framework coupling communication limits, a roofline GPU service model, stochastic latency, and utilization-aware cost across three model classes, three offloading strategies, and three communication generations, applied to New York City. Separating a reactive 100~ms budget from a 300~ms deliberative tier (presuming an onboard reactive fallback), we find three \emph{nested} binding regimes. Communication binds first in dense cells: 5G fails early, 5G-Advanced is the practical threshold for feature-level offloading, and 6G adds headroom. Compute binds next under the reactive budget: near-term VLA is latency-infeasible regardless of bandwidth, because autoregressive FP16 decode is memory-bandwidth-bound (~114 ms on 2025 hardware). Its floor clears 100 ms around 2027; 6G then admits feature-level VLA by ~2028, 5G-Advanced only at light loading and not the dense corridor, and the deliberative tier from 2026. Cost binds last: once admissible, utilization-pooled cloud GPUs undercut onboard hardware for VLA, whose baseline (up to \$8,500 per vehicle-year) is expensive and idle; feature-level offloading (S2) is where the VLA cost crossover concentrates. Latency decides which model is admissible in which year; cost decides whether it is economical.

Seonghoon Yoo, Seulhyun Kwon, Kawon Han et al.

Distributed integrated sensing and communication (D-ISAC) enables multiple spatially distributed nodes to cooperatively perform sensing and communication. However, achieving coherent cooperation across distributed nodes is challenging due to practical impairments. In particular, residual phase synchronization errors result in imperfect channel state information (CSI), while angle-of-arrival (AoA) uncertainties induce radar cross-section (RCS) variations. These impairments jointly degrade target detection performance in D-ISAC systems. To address these challenges jointly, this paper proposes a robust beamforming design for coherent D-ISAC systems. Multiple distributed nodes coordinated by a central unit (CU) jointly perform joint transmission coordinated multipoint (JT-CoMP) communication and multi-input multi-output (MIMO) radar sensing to detect a target while serving multiple user equipments (UEs). We formulate a robust beamforming problem that maximizes the expected Kullback-Leibler divergence (KLD) under statistical RCS variations while satisfying system power and per-user minimum signal-to-interference-plus-noise ratio (SINR) constraints under imperfect CSI to ensure the communication quality of service (QoS). The problem is solved using semidefinite relaxation (SDR) and successive convex approximation (SCA), and numerical results show that the proposed method achieves up to 3 dB signal-to-clutter-plus-noise ratio (SCNR) gain over the conventional beamforming schemes for target detection while maintaining the required communication QoS.