In this paper, a received signal strength (RSS) based localization technique is investigated for underwater optical wireless sensor networks (UOWSNs) where optical noise sources (e.g., sunlight, background, thermal, and dark current) and channel impairments of seawater (e.g., absorption, scattering, and turbulence) pose significant challenges. Hence, we propose a localization technique that works on the noisy ranging measurements embedded in a higher dimensional space and localize the sensor network in a low dimensional space. Once the neighborhood information is measured, a weighted network graph is constructed, which contains the one-hop neighbor distance estimations. A novel approach is developed to complete the missing distances in the kernel matrix. The output of the proposed technique is fused with Helmert transformation to refine the final location estimation with the help of anchors. The simulation results show that the root means square positioning error (RMSPE) of the proposed technique is more robust and accurate compared to baseline and manifold regularization.
|Original language||English (US)|
|Title of host publication||2018 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)|
|Publisher||Institute of Electrical and Electronics Engineers (IEEE)|
|Number of pages||5|
|State||Published - Sep 21 2018|