Wide Field-of-View Reception and Beam Direction Estimation Based on Liquid-Filled Optical Receiver

In this paper, we design a wide field-of-view (FOV) array receiver assisted by turbid liquid, which facilitates the beam direction estimation using multiple avalanche photodiodes (APDs). A back-propagation neural network (BP-NN) is proposed to estimate the beam direction based on the APD output ratios in the designed receiver. In order to reduce the number of measurement-based training samples in the neural network, we propose to incorporate the symmetry property of the receiver and sample interpolation, to expand the training samples. Experimental results demonstrate the effectiveness of the proposed estimation method for beam direction angles in both free-space and underwater environments.