Communication in maritime environments presents unique challenges often requiring the secure transfer of information over long distances in a complex dynamic environment. Here a system for generating orbital angular momentum (OAM) beams, multiplexing, transmitting, and demultiplexing using a convolutional neural network (CNN) is presented. A single input from a 1550 nm seed laser is amplified, split into four separate beams that are directed and modulated by four switches, and the resulting beams directed into phase plates to create beams carrying OAM. These four beams constitute the individual channels. The beams are passed through several optical elements, coherently combined, and transmitted to a receiver at a range of 12 m. The resulting OAM beam spatial patterns are captured using a high speed short-wave infrared detector, concurrently transmitted to a workstation for storage, and processed in real-time using a trained CNN. Results from short range and quiescent environmental state show a pattern detection accuracy of <99%.
|