Energy Optimisation through Path Selection for Underwater Wireless Sensor Networks

This paper explores energy-efficient ways of retrieving data from underwater sensor fields using autonomous underwater vehicles (AUVs). Since AUVs are battery-powered and therefore energy-constrained, their energy consumption is a critical consideration in designing underwater wireless sensor networks. The energy consumed by an AUV depends on the hydrodynamic design, speed, on-board payload and its trajectory. In this paper, we optimise the trajectory taken by the AUV deployed from a floating ship to collect data from every cluster head in an underwater sensor network and return to the ship to offload the data. The trajectory optimisation algorithm models the trajectory selection as a stochastic shortest path problem and uses reinforcement learning to select the minimum cost path, taking into account that banked turns consume more energy than straight movement. We also investigate the impact of AUV speed on its energy consumption. The results show that our algorithm improves AUV energy consumption by up to 50% compared with the Nearest Neighbour algorithm for sparse deployments.