Energy-Efficient Underwater Data Collection: A Q-Learning Based Approach

Underwater data collection is an importance part in the process of network monitoring, network management and intrusion detection. However, the limited-energy of nodes is a major challenge to collect underwater data. The solution of this problem are not only in the hands of network topology but in the hands of path of autonomous underwater vehicle (AUV). With the problem in hand, an energy-efficient data collection scheme is designed for mobile underwater network. Especially, the data collection scheme is divided into two phases, i.e., routing algorithm design for sensor nodes and path planing for AUV. With consideration of limited-energy and network robustness, Q-learning based dynamic routing algorithm is designed in the first phase to optimize the routing selection of nodes, through which a potential-game based optimal rigid graph method is proposed to balance the trade-off between the energy consumption and the network robustness. With the collected data, Q-learning based path planning strategy is proposed for AUV in the second phase to find the desired path to gather the data from data collector, then a mode-free tracking controller is developed to track the desired path accurately. Finally, the performance analysis and simulation results reveal that the proposed approach can guarantee energy-efficient and improve network stability.