Indoor Localization Using Visible Light via Two-Layer Fusion Network

In visible light communication (VLC)-based indoor localization environment, the instability and uncertainty power of emitting LEDs and other factors, as obstacles between transmitters and receivers, will lead the fluctuation of the received signal strength of receivers. To overcome the problem, this paper proposes a two-layer fusion network (TLFN) indoor localization method for VLC. The two layers in TLFN are the diverse layer and the fusion layer. In the diverse layer, TLFN obtains multiple position estimates based on the predictions of multiple fingerprints and multiple classifiers combinations. In the fusion layer, TLFN first trains and stores some weights for all grid points by minimizing the average localization errors overall fingerprints and classifiers spaces. Then, in the online phase, we propose an optimal weights searching algorithm to intelligently determine the optimal weights for fusion localization. TLFN can leverage the intrinsic supplementation among multiple position estimates to yield a higher accurate positioning result. The experiments conducted on an intensity-modulated direct detection system demonstrate that our proposed TLFN is superior to existing fusion-based approaches regardless of the instability and uncertainty power of light-emitting-diode localization environments.