ENS William A. Jarrett
Professor Svetlana Avramov-Zamurovic
Professor Joel M. Esposito
Presented at Optica Imaging and Applied Optics 2022 in Vancouver, Canada and submitted to Optics Communications
Abstract
Underwater wireless optical communication systems have garnered significant interest due to advantages in speed and security over radio-frequency and acoustic underwater transmission. One example utilizes Laguerre-Gaussian beams carrying orbital angular momentum to increase bit transfer rate by spatially multiplexing information into an alphabet of symbols, sized , where N is the number of bits encoded in each symbol. This system uses a convolutional neural network (CNN) to demultiplex the information, which has been shown to be capable of high levels of accuracy even in turbid environments or in the presence of significant optical turbulence. However, it has been observed that CNNs struggle when shown images affected by physical environmental effects (different levels of turbulence, optical magnification, beam misalignment) that were not present in the training set. In this paper we further investigate physical beam misalignment under two experimental conditions, quiescent and strong optical turbulence. Physical beam misalignment is defined as a translation of the beam at the receiver. Our optically turbulent environment is characterized using the scintillation of a Gaussian beam, and is estimated to have a refractive index structure constant value of ~ . We use two alphabet sizes, of 16 and 256 symbols, and demonstrate classification of received images in a combination of scenarios with adequate success (>90% accuracy) when training and testing under the same conditions, and using a pre-trained CNN, AlexNet. However, when training and testing under physically misaligned conditions, we demonstrate an issue with impractical rates of classification, at 5-50%. Our investigation provides CNN classification at significantly higher levels of complexity than previously seen, both in terms of experimental turbulence strength and number of classifiable symbols (alphabet size), and clearly demonstrates the importance of carefully designed experiments to aid the machine learning training process for communication systems.
Motivation
Example of Beam Misalignment
Our lightweight network is powerful and capable of learning turbulent variations quickly and effectively, but struggles when shown variations of the data in the X/Y direction
Under realistic turbulent conditions, this is caused by beam wander
Motivates investigation into methods to increase network robustness and resilience without increasing training images / symbol or changing CNN
Comparison of Misaligned Alphabets
Dataset 1 Alphabet
Dataset 2 Alphabet
Digitally Augmenting the Dataset
Random translations in training dataset +/- 5 pixels in the X/Y direction
Saw ~2 improvement in performance of the quiescent data
Saw ~4-5x improvement in the performance of the turbulent data
Conclusions
From this work, we drew out the importance of creating variation in the data used to train the network. The greater the variation, the more robust and resilient the network will be, allowing it to be used in practical applications.
For systems with limited training data, digital augmentation can be a very powerful tool.
Optica Presentation
Submitted Journal Paper