Improvement of SPGD by Gradient Descent Optimization Algorithm in Deep Learning

Abstract

The distortion caused by atmospheric turbulence can be compensated by wavefront correction, which improves the performance of free space optical vortex beam communication. In order to solve the slow convergence problem of the traditional stochastic parallel gradient descent algorithm in wavefront correction, an iterative correction algorithm based on adaptive gain factor is presented, which is combined with the Adam optimization algorithm in deep learning. The simulation verifies that the algorithm is more robust in turbulent environment.