Endogenous Encryption Scheme Driven by Multi-Dimensional Feature Fusion

Abstract

This paper proposes an endogenous encryption communication scheme based on multi-dimensional feature fusion to enhance physical-layer security. To address the limitations of conventional schemes that depend on external physical characteristics for key generation, require external key distribution and management, and rely on fixed encryption algorithms, the proposed scheme employs parallel multi-dimensional feature extraction modules to synchronously extract intrinsic features from the transmitted data, including statistical distributions, structural patterns, and complexity measures. The extracted features are then fused and mapped into endogenous keys via a feature–encryption mapping algorithm, enabling tight integration of key generation with the communication process. A 15.875 Gb/s multi-dimensional feature fusion–encrypted 16 quadrature amplitude modulation (QAM) orthogonal frequency division multiplexing (OFDM) signal transmission over a 25 km standard single-mode fiber (SSMF) is experimentally demonstrated. The experimental results show that the proposed scheme achieves reliable legitimate communication without additional bit error rate (BER) penalty, while effectively suppressing unauthorized eavesdropping, with the eavesdropper’s BER maintained at approximately 0.5. Furthermore, when only partial features are available or when a mismatched mapping algorithm is used, the BER remains close to 0.5, indicating random decoding behavior. When extracting only 11 features from the transmitted data for key generation, the key space has already reached 10154. These results verify the effectiveness of the proposed scheme in enhancing physical layer security and demonstrate its potential for highly secure and reliable optical converged transmission schemes.