Atmospheric Turbulence Strength Estimation Using Convolution Neural Network
Open Access
- 13 September 2023
- journal article
- research article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Photonics Journal
- Vol. 15 (6) , 1-7
Abstract
Laser beam transmission in atmospheric turbulence causes image distortion and affects the quality of information transmission in the field of optical communication. The strength of the atmospheric turbulence, which can be characterized by refractive index structure constant $C^{2}_{n}$ , significantly influences the properties of a laser beam. The accurate estimation of $C^{2}_{n}$ is essential for understanding the strength of turbulence. Although multilayer perceptron (MLP) and deep neural network (DNN) has been applied to estimate the atmospheric turbulence strength, the estimation accuracy is sensitive to the strength of the turbulence. In this article, we propose a method based on the convolution neural network (CNN) approach to estimate $C^{2}_{n}$ ranging from $10^{-17}$ to $10^{-13}$ $\text{m}^{-2/3}$ . We experimentally demonstrate that the correlation coefficient ( $\rm R^{2}$ ) of the model is 99.39%. The mean relative error (MRE), root mean square error (RMSE), and mean absolute error (MAE) are 0.0047, 0.0916, and 0.0684, respectively. For the turbulence strength with the same order of refractive index structure constant $C^{2}_{n}$ , the estimation accuracy of the weak turbulence is higher than that of medium and strong turbulence. Moreover, the mix training different levels of turbulence strength improves the estimation accuracy of $C^{2}_{n}$ compared to that with the same order of $C^{2}_{n}$ . Based on the high estimation accuracy of the CNN in the scheme, the proposed method will be able to provide a way of estimating the strength of atmospheric turbulence without the need for additional optical devices.Keywords
Funding Information
- Scientific Research Fund of Liaoning Provincial Education Department (LJKMZ20220620)
- 2023 Central Government Guidance for Local Science and Technology Development Funds (2023JH6/100100066)
This publication has 36 references indexed in Scilit:
- Simulating non-Kolmogorov turbulence phase screens based on equivalent structure constant and its influence on simulations of beam propagationResults in Physics, 2017
- Using an artificial neural network approach to estimate surface-layer optical turbulence at Mauna Loa, HawaiiOptics Letters, 2016
- Deep learningNature, 2015
- SLM-based laboratory simulations of Kolmogorov and non-Kolmogorov anisotropic turbulenceApplied Optics, 2015
- Measuring the intensity fluctuation of partially coherent radially polarized beams in atmospheric turbulenceOptics Express, 2014
- The Use of Weather Forecasts to Characterise Near-Surface Optical TurbulenceBoundary-Layer Meteorology, 2010
- Estimates of Cn2 from Numerical Weather Prediction Model Output and Comparison with Thermosonde DataJournal of Applied Meteorology and Climatology, 2010
- Modeling Optical Turbulence and Seeing over Mauna Kea*Journal of Applied Meteorology and Climatology, 2008
- Verification of the bulk method for calculating overwater optical turbulenceApplied Optics, 1981
- Behavior of the Refractive-Index-Structure Parameter near the Ground*Journal of the Optical Society of America, 1971