Decomposition and Efficient Learning for Adaptive Multi-Mission AUVs Collaboration in UWSNs

Underwater wireless sensor networks (UWSNs) are an important underwater monitoring and detection technology with widespread applications. However, UWSNs face severe data timeliness challenges due to the complex and dynamic underwater environment, the widely distributed sensor nodes, and the energy-constrained nature of the system. In this paper, we deploy multi-mission autonomous underwater vehicles (AUVs) with inter-AUV relay communication to enhance data collection efficiency and timeliness. Unlike traditional AUV deployment methods that rely on predefined tasks and scopes, we optimize both the task selection and trajectory planning of the AUVs to maximize the overall network’s value of information (VoI). To address this complex optimization problem, we propose a decomposition and efficient learning algorithm, which employs the QMIX algorithm for task selection and the multi-agent deep deterministic policy gradient (MADDPG) network for trajectory planning. This decomposition approach significantly improves the training efficiency and convergence speed of the algorithm. Numerical results demonstrate that our approach can effectively coordinate collaboration among multiple AUVs, leading to a substantial increase in the overall UWSNs system’s VoI.