Characterizing the spike timing of a chaotic laser by using ordinal analysis and machine learning

Area of Science:

• Nonlinear Dynamics
• Photonics
• Information Theory

Background:

• Semiconductor lasers with optical feedback exhibit complex nonlinear dynamics.
• These lasers can generate optical pulses (spikes) mimicking neural activity.
• Understanding spike timing statistics is crucial for photonic information processing.

Purpose of the Study:

• Investigate the statistical properties of laser inter-spike intervals (ISIs).
• Explore potential encoding methods for optical spikes using experimental parameters.
• Determine if ISI sequences can be differentiated based on applied sinusoidal signals.

Main Methods:

• Analysis of experimental ISI sequences from semiconductor lasers.
• Application of ordinal analysis and machine learning techniques.
• Characterization of ISI sequences using Flicker noise parameter (α) and permutation entropy.

Main Results:

• ISI sequences exhibit statistical ordinal properties similar to Flicker noise.
• The Flicker noise parameter α varies with applied sinusoidal signal parameters (DC value, frequency).
• The (α, permutation entropy) plane effectively differentiates ISI sequences from different experimental conditions.

Conclusions:

• Optical spike timing statistics in semiconductor lasers are controllable via external signals.
• Ordinal analysis and machine learning provide insights into optical spike encoding.
• The (α, permutation entropy) plane offers a robust method for analyzing and differentiating optical spike sequences.