Modelling the Refractive Index Structure Parameter: A ResNet Approach

Various atmospheric effects have a negative influence on optical signals, especially in the troposphere, which must be taken into account in free space optical (FSO) communication systems. To obtain a quantitative estimate of these effects, different mathematical models are used, often based on empirical data from around the world. The main problem with existing models is the limited accuracy, due to the different meteorological conditions at different locations on earth. We propose a new approach of modelling the refractive index structure parameter using residual neural networks (ResNets). New models, tailored to the meteorological conditions at any place on earth, can be easily created, which yields in a more accurate estimation of the refractive index profile.