A Maejo University mathematics study introduces dual-split-quaternion algebra for fractional-order memristive neural networks, proves synchronization within an initial-condition-independent fixed upper time, and uses hyperchaotic states in a 6D colour-image encryption construction. Its main contribution is theoretical; it does not certify production-grade cryptographic security.
Key findings
- The proofs yield a computable T-max independent of initial conditions under the model assumptions. Embedded simulations show DSQ-state synchronization, and the proposed image scheme reports statistical and differential-attack metrics.
Why this matters globally
The work extends hypercomplex neural-control theory to an algebra with more difficult properties than standard quaternions and may support future research in secure communication and multidimensional dynamics.
Thai researcher contribution
A Maejo University scholar authored the study and developed the mathematical framework, demonstrating Thai participation in advanced control and neural-network theory.
Limitations to consider
This is theory and simulation. Assumptions about delays, switching and controllers may not represent hardware; no external benchmark or device experiment is provided. Common image metrics do not replace cryptographic analysis against known-plaintext, chosen-ciphertext or implementation attacks.