Adaptive stochastic parallel gradient descent approach for efficient fiber coupling.

Area of Science:

• Optical Engineering
• Signal Processing
• Communication Systems

Background:

• Free-space optical communication systems require precise coupling of laser beams into single-mode fibers.
• Atmospheric turbulence degrades laser beam spatial coherence, complicating fiber coupling.
• Existing methods face challenges in achieving efficient and stable fiber coupling under turbulent conditions.

Purpose of the Study:

• To propose and evaluate a novel adaptive stochastic parallel gradient descent (ASPGD) method for efficient fiber coupling.
• To address the limitations of existing methods in handling turbulence-induced spatial coherence degradation.
• To improve the convergence speed and stability of the fiber coupling optimization process.

Main Methods:

• Formulated fiber coupling as a model-free optimization problem.
• Developed and applied the adaptive stochastic parallel gradient descent (ASPGD) algorithm.
• Integrated momentum and adaptive gain coefficient estimation into the SPGD method.

Main Results:

• The proposed ASPGD method demonstrated efficient fiber coupling.
• ASPGD reduced the number of iterations by 50% compared to the original SPGD method.
• The method maintained stability during the fiber coupling process.

Conclusions:

• ASPGD offers a significant improvement for fiber coupling in free-space optical communication.
• The enhanced SPGD method provides faster convergence and robust stability.
• This technique is crucial for overcoming atmospheric turbulence challenges in optical systems.