System-Level Secure Key Management For DNA-Based Image Cryptography Using Ephemeral ECDH and HKDF-SHA256

The increasing transmission of digital images over untrusted networks requires encryption mechanisms that combine strong statistical protection with secure key management. Although DNA-based image cryptography integrated with chaotic systems achieves high confusion and diffusion performance, most existing approaches rely on static or deterministically derived keys, lacking forward secrecy and structured session isolation. This paper proposes a system-level secure key management architecture for DNA-based image encryption. The framework integrates ephemeral Elliptic Curve Diffie–Hellman (ECDH) over Curve25519 (X25519) to establish session-specific shared secrets and employs RFC 5869-compliant HKDF-SHA256 for domain-separated subkey derivation. The derived subkeys are independently assigned to permutation, DNA rule selection, and diffusion stages, preventing key reuse and entropy overlap. Security analysis under a Dolev-Yao adversarial model demonstrates improved resistance to session compromise and public-key substitution in authenticated exchange. Experimental results confirm that the integration of structured key management preserves the statistical performance of the encryption engine, achieving NPCR above 99.6%, UACI near 33.4%, and entropy values approaching 7.99. The proposed framework enhances the practical security of DNA-based image cryptosystems by bridging statistical encryption techniques with modern secure key establishment protocols.