Task-Oriented Semantic Communication System for Maritime Patrol Surveillance

The rapid development of the shipping industry has propelled the diversified development of maritime patrol surveillance, where unmanned surface vehicle (USV) is tasked with performing autonomous and long-duration patrol missions. However, traditional communication frameworks struggle to process multimodal data, enable real-time decision-making, and meet the demands of dynamic maritime environments. To address these issues, we propose a Visual Question-Answering Maritime Semantic Communication System (VQA-MSCS), a task-oriented semantic communication framework. It integrates semantic communication to optimize operational efficiency. Specifically, real-time maritime imagery is captured and transmitted by an unmanned aerial vehicle (UAV), while text-based queries are formulated by a patrol USV based on mission requirements. The computational UAV (C-UAV) performs multimodal semantic analysis, aligning and interpreting the visual and textual data to generate precise answers that are sent back to the patrol USV. Based on the responses, the patrol USV dynamically adjusts its patrol trajectory and task execution strategy, thereby enhancing the efficiency of maritime operations. Simulation results demonstrate that communication efficiency and multimodal data accuracy are significantly improved by the proposed VQA-MSCS and multimodal data accuracy, making it a promising solution for intelligent maritime patrol surveillance.