.png){kind=icon}
.png){kind=icon}
Return
Evaporation-Duct-Aware Joint Communication–Navigation Trajectory Optimization for Multi-Waypoint Maritime Missions
.png){kind=icon}
OA
AI
H
J
Y
Y
朱
S
陶
DOI：10.1109/ojvt.2026.3716228.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Reliable long-range maritime communication remains challenging for shore-based systems due to the radio-horizon limitation and the strong spatial variability of sea-surface propagation. This paper investigates an evaporation-duct-aware multi-waypoint maritime communication mission, where a surface vessel sequentially visits prescribed task waypoints and offloads the data demand generated at each waypoint to a shore-based base station. By exploiting an evaporation-duct-aware channel gain map (CGM), we formulate a bi-objective trajectory optimization problem that jointly minimizes communication time and mission duration under waypoint-ordering, segment-wise communication-completion, and maneuverability constraints. To solve this long-horizon continuous-control problem, we develop a map-assisted Proximal Policy Optimization (PPO) framework, termed DuctMap-PPO. The proposed method adopts a Beta-policy parameterization to generate feasible bounded heading increments and designs a stage-gated reward mechanism to coordinate communication-demand completion and waypoint progression under different communication—navigation preferences. Simulation results show that CGM assistance is essential for efficient multi-waypoint maritime communication. Compared with a population-based multi-objective baseline and standard PPO, DuctMap-PPO achieves a better Pareto front at a lower computational cost. The results further demonstrate interpretable trajectory behaviors under different preference weights, useful adaptability across different evaporation duct conditions, and the effectiveness of the proposed action parameterization and reward design.
Keywords:
Maritime communication
evaporation duct
channel gain map
joint communication–navigation optimization
proximal policy optimization
reinforcement learning
Journal
I
IF:
4.8
Papers:
493
Citations:
987
Organization
Cited Papers