Increasing the Efficiency of Free-Space Optical Communication Systems Using Neural Networks

Abstract

Weather conditions, such as dust, rain, and fog, significantly reduce the power of the laser used to transmit optical signals by altering the propagation characteristics of the light path. This degradation impacts the reliability and quality of free-space optical communication systems. This research aims primarily to develop a simple and intelligent solution to compensate for this atmospheric attenuation using a two-layer artificial neural network (ANN). Unlike conventional systems that rely on complex optical compensation mechanisms, the proposed ANN-based approach provides an efficient, less equipment-intensive alternative.

The process entails simulating various disturbance circumstances with MATLAB to simulate the behavior of genuine free-space channels. Experimenting with different attenuation levels in dust, rain, and fog conditions yields a dataset. This data is then used to train and test the neural network, which learns to forecast how much laser power is needed to maintain and improve signal integrity. The simulation program uses a 1550 nm wavelength and a 10 dB power factor to ensure accurate modeling. Key assumptions include the network's capacity to generalize attenuation patterns and recover the best power for transmission.