Chaos key enhanced physical layer secure transmission method based on the convolutional long short-term memory neural...

Area of Science:

• Optical Communications
• Cryptography
• Machine Learning

Background:

• Traditional chaotic encryption methods face security risks due to chaotic dynamics degradation.
• Enhancing key space is crucial for next-generation secure transmission systems.

Purpose of the Study:

• To propose a key-enhanced chaotic sequence training method using CLSTM-NN for secure transmission.
• To address security vulnerabilities in conventional chaotic encryption.
• To improve the key space and demonstrate feasibility in optical communication systems.

Main Methods:

• Training a key-enhanced chaotic sequence using a convolutional long short-term memory neural network (CLSTM-NN).
• Applying the method to orthogonal chirp division multiplexing (OCDM) systems.
• Conducting transmission experiments with encrypted 16-QAM OCDM signals at 53.25 Gb/s over 7-core optical fiber.

Main Results:

• The proposed method expands the key space from 10^36 to 10^241 compared to traditional chaotic models.
• Key enhancement increased the system's overall key count from 18 to 10^5.
• Bit error rate (BER) performance showed no significant difference compared to traditional methods, with a maximum performance difference not exceeding 1 dBm.

Conclusions:

• The CLSTM-NN based key-enhanced chaotic sequence method is feasible for secure optical transmission.
• The approach offers a substantial improvement in key space, enhancing security.
• This provides a promising direction for future secure communication technologies.