Time delay estimation from the time series for optical chaos systems using deep learning.

Area of Science:

• Physics
• Engineering
• Computer Science

Background:

• Optical chaos systems are crucial for secure communication, with time delay signatures (TDS) being key security factors.
• Existing methods like autocorrelation function (ACF) and delayed mutual information (DMI) struggle with high nonlinearity and specific concealment strategies.
• Previous deep learning approaches required knowledge of the chaos system's structure.

Purpose of the Study:

• To develop a model-free method for extracting time delay signatures (TDS) from optical chaos systems.
• To enhance the security of optical chaos communication by effectively identifying concealed TDS.
• To propose a blind identification method robust to system nonlinearity and noise.

Main Methods:

• A blind identification method based on a long short-term memory neural network (LSTM-NN) model is formulated.
• The method is validated on opto-electronic oscillator (OEO) chaos systems and laser chaos systems.
• The approach is tested against security-enhanced chaotic systems and additive noise.

Main Results:

• The proposed LSTM-NN method successfully extracts TDS without prior knowledge of system dynamics.
• The method demonstrates high tolerance to additive noise.
• It requires less data compared to existing TDS extraction techniques.

Conclusions:

• The LSTM-NN based model-free approach offers a robust and efficient solution for TDS extraction in optical chaos systems.
• This method significantly improves the security analysis and potential implementation of chaos-based secure communication.
• The technique is adaptable to various optical chaos systems, including those with enhanced security features.