Protected memristive implementations of cryptographic functions

Area of Science:

• Computer Science
• Electrical Engineering
• Cryptography

Background:

• Von Neumann architectures face performance limitations due to the memory wall.
• Memristive devices enable in-memory processing, offering a novel approach beyond traditional CMOS technology.
• Security and cryptographic functions are critical areas for exploring memristor applications.

Purpose of the Study:

• To investigate memristor-oriented countermeasures against cryptographic vulnerabilities.
• To analyze the effectiveness of hiding and masking techniques in memristive cryptographic implementations.
• To evaluate the feasibility of memristor-based security solutions for future computing platforms.

Main Methods:

• Review of memristive crossbar configurations for cryptographic functions.
• Implementation of a four-bit S-box using memristor technology.
• Analysis of memristor-oriented hiding and masked dual-rail pre-charge logic (MDPL) masking techniques on a self-rectifying passive crossbar.

Main Results:

• Memristor-oriented hiding techniques demonstrate significant efficacy.
• Memristor-oriented MDPL masking methods show limitations, particularly concerning power consumption profiles of different input data.
• The xor4Sbox with CRS-based MDPL masking failed stochastic tests due to power consumption discrepancies, despite passing t-tests.

Conclusions:

• Memristive technology presents viable options for in-memory computing and cryptographic functions.
• Further research is needed to optimize memristor-based masking techniques for robust security.
• This work contributes to understanding the challenges and opportunities of memristors in secure computing.