Known-plaintext attack to optical encryption systems with space and polarization encoding.

Area of Science:

• Cryptography
• Optical Engineering
• Information Security

Background:

• Space-based optical encryption (SBOE) and double random polarization encoding (DRPO) are advanced optical cryptosystems.
• These systems were thought to be highly secure against common attacks.
• Known-plaintext attacks (KPA) on SBOE and DRPO have been rarely explored.

Purpose of the Study:

• To investigate and develop a novel known-plaintext attack (KPA) against SBOE and DRPO.
• To demonstrate the security vulnerabilities of these advanced optical encryption methods.
• To propose a method for cracking linear optical cryptosystems.

Main Methods:

• A matrix regression approach using training samples to model plaintext-ciphertext relationships.
• Directly modeling the relationship via a complex-amplitude weighting matrix.
• Optimizing the matrix using a gradient descent algorithm for efficient KPA implementation.

Main Results:

• The proposed matrix regression KPA effectively cracks SBOE and DRPO.
• The attack can be performed without recovering the exact encryption key.
• The method is simpler than deep learning approaches for cryptanalysis.

Conclusions:

• SBOE and DRPO possess significant security flaws exploitable by the proposed KPA.
• The matrix regression method provides a viable approach for cryptanalyzing linear optical systems.
• This research highlights the need for re-evaluating the security of current optical cryptosystems.