A Novel Hybrid Communication Policy Using Network Coding Based Post-Quantum Cryptography and Adaptive Neuro Fuzzy Inference System

Abstract

This paper presents a novel hybrid communication policy developed using Hybrid Universal Network Coding Cryptosystem (HUNCC), which can perform post-quantum cryptography at a high information rate. This hybrid scheme (HUNCC) is based on the notion of individual secrecy and computational complexity, where we assume that if the eavesdropper has access to a small part of the communication link (ciphertext) between the sender and receiver, individual secrecy will ensure secure communication. However, If the eavesdropper can access the full link (ciphertext), then computational complexity can guarantee post-quantum security. We extend our study by applying fuzzy logic to HUNCC, helping us gain more insights. Specifically, we propose a novel Adaptive Neuro Fuzzy Inference System that provides a security metric from the input attributes, namely the physical layer (PHY) security attribute and cooperative sensing (CS) security attribute obtained from trusted sources. We assume such attributes are available for the ANFIS to evaluate a security metric that will decide on whether (i). to implement the cryptosystem over a single link in a multi-link system (HUNCC) or (ii). to completely use the notion of individual secrecy alone to encode the system, provided the output from ANFIS guarantees good security. This method can help switch between classical and quantum-secure communication, improving the effective throughput. Using HUNCC with a modified McEliece Cryptosystem that uses Hamming code, we obtain an information rate of 0.9507 bits, illustrating the increase in throughput obtained using the modified HUNCC model when compared to existing PQC techniques. Furthermore, we introduce a novel concept aimed at achieving even higher information rates. This approach relies on the security metric derived from ANFIS and involves dynamic switching between quantum and classical communication, guided by security risk factors. As a result, we achieve an information rate of up to 0.98 bits.