Orbital angular momentum distortion correction and mode identification using a multitask neural network.

Area of Science:

• Optics and Photonics
• Machine Learning
• Ocean Engineering

Background:

• Orbital angular momentum (OAM) vortex beams enhance underwater wireless optical communication (UWOC) capacity.
• Oceanic turbulence (OT) causes beam distortion, challenging OAM transmission and identification.

Purpose of the Study:

• To propose a multitask neural network (MTNN) for distortion correction and OAM mode identification in UWOC systems.
• To mitigate the impact of oceanic turbulence (OT) on OAM beam quality and recognition.

Main Methods:

• A multitask neural network (MTNN) was developed for feature extraction from distorted OAM intensity distributions.
• The MTNN model incorporates two output branches for predicting OT phase screens and OAM modes.
• The proposed technique processes distorted OAM intensity data to correct for channel-induced aberrations.

Main Results:

• The MTNN effectively eliminated OT, corrected OAM distortion, and accurately identified OAM modes.
• Mode purity and identification accuracy of OAM beams were significantly enhanced post-correction.
• The MTNN demonstrated superior anti-turbulence performance compared to conventional convolutional neural networks (CNNs) across varying OT strengths.

Conclusions:

• The proposed MTNN-based scheme offers a robust solution for high-quality OAM transmission in turbulent underwater environments.
• This technology enables innovative advancements in high-performance UWOC systems.
• The technique successfully addresses challenges in OAM transmission and identification within UWOC systems.