De-multiplexing vortex modes in optical communications using transport-based pattern recognition.

Area of Science:

• Optical Communications
• Signal Processing
• Machine Learning

Background:

• Free space optical (FSO) communication systems use orbital angular momentum (OAM) beams for increased channel capacity.
• Atmospheric turbulence distorts OAM beams, complicating signal demultiplexing and limiting communication performance.
• Deep learning (DL) methods have been applied to OAM beam demultiplexing as an image classification problem.

Purpose of the Study:

• To introduce a novel, computationally efficient shallow learning method for decoding OAM beams in turbulent FSO communication channels.
• To leverage a new theoretical framework linking image turbulence to photon transport via the continuity equation.

Main Methods:

• Developed a shallow learning technique based on the continuity equation to model and mitigate turbulence effects in OAM beam demultiplexing.
• Compared the performance of the proposed shallow learning method against deep convolutional neural networks (CNNs) using bit error ratio (BER) as a metric.

Main Results:

• The shallow learning method achieved comparable classification accuracy (BER) to deep learning approaches.
• The new method demonstrated a significant reduction in computational cost, requiring only 1/90th of the resources used by deep CNNs.
• This computational efficiency enables higher achievable bit rates in FSO communication systems.

Conclusions:

• Shallow learning offers a viable and efficient alternative to deep learning for OAM beam demultiplexing in turbulent FSO channels.
• The proposed method provides a practical solution for enhancing the performance and data rates of OAM-based FSO communication systems.
• Reduced computational complexity facilitates the deployment of advanced optical communication technologies.