Abstract
<title>Abstract</title> <p> <bold>This paper presents an intelligent adaptive hybrid cryptographic framework that combines the Unimodular Hill Cipher, RSA-OAEP, and Logistic Chaotic Map-based key generation to provide secure protection for universal binary files. In contrast to traditional encryption schemes that employ fixed encryption strategies, the proposed framework utilizes an AI-driven adaptive decision mechanism to determine the optimal encryption allocation according to the statistical characteristics of the input file. The decision process considers several features, including file entropy, byte frequency distribution, file size, standard deviation, and the number of distinct byte values. The adaptive strategy encrypts most bulk data using the computationally efficient Unimodular Hill Cipher, while sensitive data segments are secured with RSA-OAEP to strengthen asymmetric protection. To enhance key randomness and unpredictability, Logistic Chaotic Maps are employed to generate dynamic unimodular matrices that exhibit high sensitivity to initial conditions. The proposed approach was evaluated using multiple binary PDF datasets under various matrix dimensions and password configurations. Experimental results demonstrate excellent cryptographic performance, achieving a ciphertext entropy of 7.999891, an almost zero plaintext–ciphertext correlation coefficient of −0.000152, and a substantial reduction in the Chi-Square statistic from 8159.84 to 269.85, indicating improved statistical uniformity and diffusion characteristics. Moreover, identical SHA-256 hash values between the original and decrypted files verify complete and lossless data recovery. Overall, the experimental findings demonstrate that the proposed adaptive hybrid cryptosystem offers robust security, high computational efficiency, dynamic adaptability, and broad compatibility for protecting heterogeneous binary files across diverse digital environments.</bold> </p>