Message Group Optimization in Kyber-512 Post-Quantum Cryptography: Comparing Proposed Static Encoding With Huffman Code, Base-64, and ASCII Methods
Kritsanapong Somsuk · IEEE Access · 2025
Post-Quantum Cryptography (PQC) indicates a significant progression in cryptographic research. Kyber-512, a principal algorithm in the Kyber family, has attained international standard certification in PQC. A primary limitation of Kyber-512 is the requirement to convert plaintext into a polynomial equation of degree 256 before the encapsulation process is applied to satisfy the parameter specifications. Then, greater computational resources are required, especially when evaluating large messages that require segmentation into smaller parts. This paper proposes techniques to reduce the bit-length per character, hence minimizing the number of segments requiring for encapsulation and decapsulation processes. The method utilizes a customized mapping table allocating a fixed-length 6-bit code to each fundamental character, including uppercase English letters, digits, and certain symbols, amounting to a total of 54 characters. In fact, two techniques are presented for integration with the mapping table and Kyber-512. The first method divides the characters into groups of 42, because it is the largest size that remains inside 256 bits. This method generates a 4-bit waste in each group. Therefore, the second method is implemented to resolve this problem by shifting the bits across groups. Furthermore, Pre-Check-Tailing (PRT) and Post-Check-Tailing (PST) are suggested to reduce the mistakes induced by padding on the plaintext’s tail segment. The evaluation includes a comparative examination of ASCII Code, Base-64 and Huffman Code. Experimental results indicate that Huffman Code achieves excellent data reduction and rapid processing speed, because of the frequency-based compression methodology. However, the dependence on providing a Huffman tree with the message presents possible security vulnerabilities by disclosing statistical characteristics of the original plaintext. Although a Huffman tree can also be encapsulated together with the compressed plaintext, this requires a double-encryption structure, which increases computational overhead and contradicts the goal of minimizing processing requirements in PQC. In fact, the proposed methods ensure consistent security attributes and demonstrate a significant reduction in message segmentation related to ASCII Code and Base-64. It has significant promise for application in resource-constrained situations, such as IoT devices and smartphones, providing diminished processing requirements without risking security.