Queue-Aware Graph Reinforcement Learning for UAV-ISAC-Assisted Maritime Data Collection

Jul 16, 2026

Upload by Logic

This paper studies high-altitude platform (HAP)-assisted sparse cooperative integrated sensing and communication (ISAC) for UAV-enabled ocean monitoring. A fleet of rotary-wing UAVs senses drifting buoys, collects monitoring data, and forwards local posterior estimates to a HAP for fusion and sparse cooperation control. The model incorporates spatially correlated sea-patch fields, patch-aware buoy dynamics, RCS-and clutter-aware echo sensing, fused posterior Cramér-Rao bounds, and propulsion-energy-limited mobility. A queue-weighted buffered-collection long-horizon Markov decision process and a structured feasible-association graph-MARL framework improve queue-weighted collection utility and transfer to larger networks.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages17

Reading Time43 min

Introduction

Related Works

UAV-Enabled ISAC

arXiv :2607 .00324v1 [ ee ss . SY] 1 Jul 2026 Queue-Aware Graph Reinforcement Learning for UAV-ISAC-Assisted Maritime Data Collection Bohan Li, Member, IEEE, Min Ye, Haochen Liu, Yongkang Gong, Member, IEEE, Ning Gao, Member, IEEE, Jie Nie, Member, IEEE, Pei Xiao, Senior Member, IEEE, Xiuzhen Cheng, Fellow, IEEE Abstract—This paper studies high-altitude platform (HAP)-assisted sparse cooperative integrated sensing and communication (ISAC) for UAV-enabled ocean monitoring. A fleet of rotary-wing UAVs senses drifting buoys, collects their monitoring data, and reports local posterior estimates to a HAP that performs fusion and sparse cooperation control. The model explicitly accounts for a spatially correlated sea-patch field, patch-aware buoy dynamics, RCS-and clutter-aware echo sensing, fused posterior Cramér-Rao bounds (PCRBs), and propulsion-energy-limited UAV mobility. The long-horizon objective is cast as a queue-weighted buffered-collection Markov decision process rather than instantaneous throughput, where each buoy maintains a backlog of buffered observations. The resulting long-horizon design is formulated as a mixed discrete-continuous problem with sensing, communication, mobility, safety, buffered-collection, and onboard-energy constraints. To address the combinatorial association component without replacing learning by a deterministic optimizer, we propose a structured feasible-association graph-MARL framework. A heterogeneous graph encoder produces candidate-edge logits, and a masked sequential b-matching policy samples legal UAV-buoy associations while exactly satisfying UAV-load and buoy-cluster constraints. A MAPPO-style training procedure, an independent queue-state value critic, and a consistency-verification protocol are then specified to support reproducible training. Simulation results on congested maritime scenarios show that the proposed policy improves the cumulative queue-weighted collection utility by about 106% over the rate-driven deterministic decoder, maintains a large margin across sea-state sweeps and medium-to-heavy traffic loads, and transfers to larger networks without fine-tuning. Index Terms—Integrated sensing and communication, maritime monitoring, UAV networks, graph neural networks, multi-agent reinforcement learning. ~~ ~~ ✦ ~~ ~~ 1 INTRODUCTION Ocean monitoring networks underpin environmental sensing, maritime safety, and the blue economy, relying on dense fleets of low-cost surface buoys and sensors to report oceanographic observations over wide, infrastructure-sparse sea areas [1] . The lack of reliable offshore backhaul and the low transmit power of such terminals make timely data delivery difficult, which motivates aerial relays that can approach the terminals on demand [2, 3] . Unmanned aerial vehicles (UAVs) are especially attractive here because their controllable mobility lets them establish strong lineof-sight uplinks to dispersed buoys and collect their buffered observations efficiently. In parallel, integrated sensing and communication (ISAC), which reuses a common waveform and aperture for both functions, has emerged as a key sixth-generation (6G) capability [4, 5] . Endowing maritime UAVs with ISAC lets a single platform localize drifting buoys and collect their data within one resource budget, a synergy well matched to the dual sensing-andcollection nature of ocean monitoring [6] . B. Li, M. Ye and J. Nie are with the Faculty of Information Science and Engineering, the Engineering Research Center of Advanced Marine Physical Instruments and Equipment (Ministry of Education), and Qingdao Key Lab oratory of Optics and Optoelectronics, Ocean University of China, Qingdao 266100, China ([emails: {bohan.li](emails: {bohan.li), yemin, [niejie}@ouc.edu.cn](niejie}@ouc.edu.cn)). H. Liu is with the School of Electronics and Information, Northwestern Poly technical University, Xi’an 710019, China (email: [haochenliu@nwpu.edu.cn](haochenliu@nwpu.edu.cn)). Y. Gong a

FAQs

What problem does the paper target in UAV-assisted maritime monitoring?

It targets long-horizon UAV-ISAC design where sensing accuracy, uplink rate, propulsion energy, and UAV-buoy association are tightly coupled under drift, clutter, and limited onboard energy, making myopic throughput-based scheduling insufficient.

How is the long-horizon objective defined?

The objective is formulated as a queue-weighted buffered-collection Markov decision process, where each buoy maintains a backlog of buffered observations rather than optimizing instantaneous throughput.

What method does the paper use to handle the association decision without relying on a deterministic optimizer?

It proposes a structured feasible-association graph-MARL framework with a heterogeneous graph encoder that produces candidate-edge logits and a masked sequential b-matching policy to sample legal UAV-buoy associations while satisfying UAV-load and buoy-cluster constraints.

Jul 16, 2026

Upload by Logic