M² factor estimation in few-mode fibers based on a shallow neural network

Area of Science:

• Optical Engineering
• Computational Physics

Background:

• Accurate M² factor estimation is crucial for laser beam quality assessment.
• Existing methods can be time-consuming or costly, limiting real-time applications.

Purpose of the Study:

• To develop a high-accuracy, high-speed, and low-cost M² factor estimation method for few-mode fibers.
• To utilize a shallow neural network for direct M² factor estimation from near-field images.

Main Methods:

• A dimensionality reduction technique transforms 2D near-field images into 1D vectors.
• A shallow neural network with two hidden layers is trained for M² factor estimation.
• The method is validated through both simulations and experimental measurements.

Main Results:

• Simulations show mean estimation errors below 3% for up to 10 modes.
• The method estimates M² factors for 10,000 samples in approximately 0.16 seconds.
• Experimental results with a 3-mode fiber yield a mean estimation error of 0.86%.

Conclusions:

• The proposed shallow neural network method offers a fast, accurate, and cost-effective solution for M² factor estimation in few-mode fibers.
• The technique demonstrates significant potential for real-time laser characterization.
• The underlying strategies are adaptable to other laser characterization tasks.