Jointly recognizing OAM mode and compensating wavefront distortion using one convolutional neural network.

Area of Science:

• Optics and Photonics
• Artificial Intelligence
• Signal Processing

Background:

• Orbital angular momentum (OAM) multiplexing is a key technology for increasing communication capacity.
• Accurate OAM mode recognition is crucial for the performance of OAM-based systems.
• Traditional adaptive optics (AO) systems often require complex hardware and additional components.

Purpose of the Study:

• To propose a novel and concise method for recognizing OAM modes.
• To enhance the accuracy of OAM recognition in the presence of atmospheric turbulence.
• To integrate wavefront sensor-less (WFS-less) AO capabilities into the OAM recognition process.

Main Methods:

• A jointly trained convolutional neural network (CNN) with a shared model backbone was developed.
• The CNN implicitly incorporates AO correction by utilizing additional mode information during offline training.
• The system structure is simplified, eliminating the need for separate AO components.

Main Results:

• Numerical simulations demonstrated significant improvements in OAM recognition accuracy under various turbulence conditions.
• The proposed method achieved performance comparable to traditional AO-combined approaches.
• The integrated CNN-based AO approach proved effective in enhancing system robustness.

Conclusions:

• The proposed method offers an efficient and accurate solution for OAM recognition.
• Combining mode recognition with WFS-less AO via a shared CNN backbone simplifies system design.
• This approach shows great potential for practical applications in free-space optical communication and other OAM-based technologies.