On the noise effect of fingerprinting-based positioning error in underwater visible light networks

This paper assesses the performance of a localization technique for underwater visible light networks. The proposed approach is based on a fingerprinting technique, collecting the channel impulse responses from different wireless optical signals in the visible range. A local database related to the power level distribution within a maritime environment is built and exploited to estimate user position, e.g., a diver moving in a given space for underwater fish monitoring. In this paper, we investigate on the noise effect on the localization accuracy in underwater scenarios and for different water turbidity coefficient and we demonstrate that the estimation error suffers on variable channel impulse responses. Different configuration parameters and environmental scenarios have been considered, showing that the LED transmitter deployment can be effective in the localization estimation. A comparison of the proposed localization approach to the traditional triangulation method has been finally carried out, showing the effectiveness of the fingerprinting-based solution for a lower number of LED transmitters.