Performance Evaluation of Underwater Visible Light Positioning Algorithms using Realistic Propagation Model

The internet of underwater things is a promising paradigm for many ocean applications, which can be based on visible light communication. A few papers study the positioning of autonomous underwater vehicles or divers using underwater visible light communication. Two algorithms have been proposed recently, using a few transmitters placed on the sea bed. They build a database of channel's gains for different regular positions in the considered underwater environment using the Beer-Lambert model, a very simple light propagation model. They were validated using pseudo-measurements of channel's gains at unknown positions using the same simple propagation model. This paper investigates whether these two algorithms are still valid using a more realistic light propagation model based on Monte-Carlo integration to approximate the measured channel's gain that can be expected under real-world conditions. It first evaluates the accuracy of these algorithms using this Monte-Carlo model for estimating the pseudo-measurements while keeping the database based on Beer-Lambert. Furthermore, it does the same study using Monte-Carlo model for both pseudo-measurements and database construction. The conclusions are that these two positioning algorithms are still valid when considering a more realistic propagation model, for instance in harbor water with first algorithm from 1.49m to 0.56m in mean positioning error. Moreover, they need such a model for the construction of the database to be used with real measurements, as the mean error grows up to 4.55m with Beer-Lambert for database construction and realistic model for pseudo-measurements.