High-security probabilistic shaping method via chaotic tagging

作者：Qingwei Wang, Bo Liu, Jianxin Ren, Yaya Mao, Shuaidong Chen, xiumin song, Xiangyu Wu, JiYuan Wang, zining pan, Xian Lu, Mingqian Liu · 发表于：Optics Letters · 年份：2026 · DOI：10.1364/ol.588892 · 研究领域：Chaos-based Image/Signal Encryption、Chaos control and synchronization、Cryptographic Implementations and Security

This paper proposes a high-security probabilistic shaping 16-7QAM mapping method via constellation-level chaotic tagging. A novel, to the best of our knowledge, constellation-level chaotic tagging rule is devised that re-maps the original 16 signal points, each encoded by 4 bits, onto predefined 3-bit modulation symbols. Simultaneously, a Lorenz chaotic system is employed to generate a chaotic sequence, from which a 2-bit tag corresponding to each 3-bit modulation symbol is extracted to form the constellation-level chaotic tag. Subsequently, each signal point is combined with a 2-bit tag generated by a Lorenz chaotic system to form the final output consisting of 7 signal points. Experimental validation over a seven-core fiber transmission system shows that, at a bit error rate of 3.8 × 10 −3 , the proposed probabilistic shaping 16-7QAM constellation achieves a 0.49 dB improvement in receiver sensitivity compared with conventional 16-QAM. Moreover, the encryption scheme has almost no impact on the transmission performance of the probabilistic shaping signal itself. Additionally, the key space of the scheme reaches 10 105 , effectively ensuring physical layer security. The proposed method exhibits robust performance and strong security, enhancing both the transmission performance and security of the system.