Abstract
Free-space optical communications systems suffer from turbulent propagation of light through the atmosphere, attenuation, and receiver detector noise. These effects degrade the quality of the received state, increase cross-talk, and decrease symbol classification accuracy. We develop a state-of-the-art generative neural network (GNN) and convolutional neural network (CNN) system in combination, and demonstrate its efficacy in simulated and experimental communications settings. Experimentally, the GNN system corrects for distortion and reduces detector noise, resulting in nearly identical-to-desired mode profiles at the receiver, requiring no feedback or adaptive optics. Classification accuracy is significantly improved when these generated modes are demodulated using a CNN that is pre-trained with undistorted modes. Using the GNN and CNN system exclusively pre-trained with simulated optical profiles, we show a reduction in cross-talk between experimentally-detected noisy/distorted modes at the receiver. This scalable scheme may provide a concrete and effective demodulation technique for establishing long-range classical and quantum communication links.
Highlights
Free-space optical communications systems suffer from turbulent propagation of light through the atmosphere, attenuation, and receiver detector noise
In order to gain insight into the dependence of our generative neural network (GNN)’s performance on orbital angular momentum (OAM) value, we show the received and corresponding corrected mean squared error (MSE) values for each ±l from 0 to ±10 with a fixed C2n of 51.2 × 10−11 m−2/3 in the inset of Fig. 4b
We have developed a GNN and convolutional neural network (CNN) system that efficiently improves received signals in a free-space optical (FSO) communication scheme that have been severely deteriorated due to the effects of turbulence, attenuation, and detector noise
Summary
Free-space optical communications systems suffer from turbulent propagation of light through the atmosphere, attenuation, and receiver detector noise. Using the GNN and CNN system exclusively pre-trained with simulated optical profiles, we show a reduction in cross-talk between experimentallydetected noisy/distorted modes at the receiver This scalable scheme may provide a concrete and effective demodulation technique for establishing long-range classical and quantum communication links. We expand significantly upon these works and develop a communication scheme using a generative machine learning approach and demonstrate its robustness and ability to significantly mitigate the effects of turbulence, attenuation, and noise on the accuracy in both simulated and experimental communication settings This receiver-end system is shown to be effective for a wide range of turbulence and detector noise strengths and requires no feedback to the transmitter of the communication link or any adaptive optics components. Our network architecture is portable, cost-effective, and can be pre-trained before performing the communications, which circumvents the general technical issues that are present in adaptive optics
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
