Hu Ying-Ying, Lin Xing, Wang Shuang, Geng Jia-Qi, Yin Zhen-Qiang, Chen Wei, He De-Yong, Huang Wei, Xu Bing-Jie, Guo Guang-Can, Han Zheng-Fu
Opt Lett. 2020 Nov 1;45(21):6038-6041. doi: 10.1364/OL.409187.
We investigate quantum random number generation based on backward spontaneous Raman scattering in standard single-mode fiber, where the randomness of photon wavelength superposition and arrival time is simultaneously utilized. The experiment uses four avalanche photodiodes working in gated Geiger mode to detect backward Raman scattering photons from four different wavelength channels and a time-to-digital converter placed behind the detectors to record their arrival time. Both information of the wavelength and arrival time interval of photons from different channels are applied to generate random bits. Due to the independence of these two entropy sources, the random number resource of the present system is fully utilized. Five-bit raw data can be obtained for every effective click, which contains 2.87-bit min-entropy. To obtain the optimal generation rate of random bits, appropriate pump power and fiber length are adopted. The post-processing method by the SHA-256 hashing algorithm is used to remove the bias of the raw data, after which the final random bit sequences pass the NIST statistical test.
我们研究了基于标准单模光纤中后向自发拉曼散射的量子随机数生成，其中同时利用了光子波长叠加和到达时间的随机性。该实验使用四个工作在门控盖革模式下的雪崩光电二极管来检测来自四个不同波长通道的后向拉曼散射光子，并在探测器后面放置一个时间数字转换器来记录它们的到达时间。来自不同通道的光子的波长和到达时间间隔信息都被用于生成随机比特。由于这两个熵源的独立性，本系统的随机数资源得到了充分利用。每次有效点击可获得五位原始数据，其包含2.87位最小熵。为了获得随机比特的最佳生成速率，采用了合适的泵浦功率和光纤长度。使用SHA-256哈希算法的后处理方法来消除原始数据的偏差，之后最终的随机比特序列通过了美国国家标准与技术研究院（NIST）的统计测试。
