Authenticated Content-Aware Image Encryption Using HMAC-SHA-256, Dynamic DNA Coding, and a 2D-LSCM Hyperchaotic Map

Chaos-based ciphers are widely proposed for sensitive images on 5G and Internet of Things links. Three faults recur: a key stream derived from the image that the receiver cannot reproduce, chaos that is claimed but not verified, and ciphertext that is not authenticated. This article presents an image cipher that removes all three. A shared secret key, a fresh per-image random value, and a digest of the image are combined through a keyed hash to derive every parameter; the digest travels masked, so the receiver rebuilds the key stream exactly and the header reveals nothing about the image. A two-dimensional chaotic map with a measured operating range and arithmetic that every platform reproduces bit-for-bit drives a pixel permutation and a DNA coding whose rules change from byte to byte. A tag over header and ciphertext is checked before decryption, and a standard ChaCha20 stream can replace the chaotic stage. On the Kodak suite the ciphertexts reach 7.9995 bits of entropy per byte, pixel correlation below 0.01, and differential statistics inside their acceptance bands; on 3330 further remote sensing, medical, and mixed content images the tests reject at their nominal rates, and NIST SP 800-22 passes under five independent keys. The security evidence is empirical and design-level; standardized authenticated encryption remains preferable where a formal proof is required.