Multi-Criteria Security and Attack Resistance Analysis of Matrix-Based Encryption Scheme

Abstract

In this paper, we analyze the security of a novel matrix-based encryption algorithm by systematically evaluating its resistance to five major types of attacks: brute-force attacks, statistical analysis, differential cryptanalysis, linear cryptanalysis, and chosen-ciphertext attack (CCA) oracle tests. The algorithm is based on text-derived keys, reversible matrix transformations, and bit-level operations to provide cryptographic security. To evaluate the security features, we use seven different metrics: entropy visualization, ECB Detector, Bit Balance, Runs Test, Avalanche Effect, Compression Test, and LZ Complexity. We evaluate each metric before and after applying each type of attack and provide a comparison to understand the impact of different types of attacks on the ciphertext. In experimental tests, we find that the algorithm shows an average resistance score of 60.0% for all types of attack with statistical randomness (high entropy) and diffusion properties (strong avalanche effect). The metric analysis shows that although the algorithm is compromised in key space size and deterministic key derivation, it exhibits good statistical properties that are resistant to pattern-based and frequency-analysis attacks. The results show the effectiveness of the seven-metric framework in providing a multi-dimensional security assessment and also point to areas for improvement in key management and non-linear component integration.