基于激光相位波动的高速量子随机数发生器

发布时间：2018-03-30 19:11

本文选题：量子随机数　切入点：后处理　出处：《中国科学技术大学》2017年博士论文

【摘要】：本文针对包括量子密钥分发等现代安全通信协议在内的科技和工业领域对高质量、高产生速度的随机数的需求,研制了小型集成化的高速量子随机数发生器样机。所研制发生器的量子随机数的实时产生速度达到了 3.2 Gbps,并能通过NIST随机性检验,符合真随机数的基本特征。本文对不同的随机数发生方案进行了比较,最终选择了基于激光相位波动的量子随机数发生方案,同时通过建模的方法对这种方案的可行性进行了理论分析,并以实验的方法确定了系统的最优工作点。在基于激光相位波动的量子随机数发生方案中,量子相位波动作为量子信号源,其随机性来源于激光的自发辐射。干涉仪可以将不可直接测量的相位波动转换为可测量的光强波动,然后通过高速光电探测器转换为随机波动的电信号,进而进行采集量化获得原始的量子随机数。除此之外,本文还设计了用于测试目的的内存条模块以及光纤、千兆位以太网、USB 2.0等数据传输接口以满足不同的应用需求,其中通过光纤传输通道量子随机数的有效输出速率为3.2 Gbps。本文在研制基于激光相位波动的高速量子随机数发生器过程中,针对安全性经信息论证明的后处理对原始随机数处理速度慢的问题,提出了一种在FPGA中实现的高速后处理算法,实时后处理速度达到了 3.36Gbps,解决了技术瓶颈。受到经典噪声和实际器件的非理想性等因素的不利影响,原始随机数往往并非理想的真随机数,需要经过后处理才能获得符合真随机数特征的量子随机数。而现有的安全性经信息论证明的后处理过程计算量大,导致处理速度缓慢,所以对原始随机数一般只能进行离线的而非高速实时的后处理,这严重限制了量子随机数的实时产生速度。针对这一问题,我们通过认真研究基于大型Toeplitz矩阵后处理的计算特点,提出了一种基于矩阵分解的并行算法,显著降低了该算法所需的硬件逻辑资源,成功地在FPGA中实现了基于大型Toeplitz矩阵的高速后处理运算,实时后处理速度达到了 3.36 Gbps。我们对经过后处理的量子随机数进行了统计均匀性、自相关性、NIST检验等随机性测试,测试结果显示量子随机数符合真随机数的基本特征,说明后处理算法有效。这种在FPGA中并行实现的高速后处理算法解决了后处理速度与原始随机数发生速度严重不匹配问题,使基于激光相位波动的量子随机数发生器能够真正产生高速的量子随机数。本文的主要创新点和创新成果如下:1)针对包括量子密钥分发等现代安全通信协议在内的科技和工业领域对高质量、高产生速度的随机数的需求,成功研制了小型集成化的高速量子随机数发生器样机。所研制发生器的量子随机数的实时产生速度达到了 3.2 Gbps,并能通过NIST随机性检验,符合真随机数的基本特征。2)针对安全性经信息论证明的后处理对原始随机数处理速度慢的问题,提出了基于矩阵分解的高速后处理算法,显著降低了该算法所需的逻辑资源,成功地在FPGA中实现了大型Toeplitz矩阵后处理运算,解决了高速随机数后处理速度上的技术瓶颈,并形成了专利技术。
[Abstract]:This paper includes quantum key distribution and so on high quality modern security communication protocol, science and technology fields, high speed random number generation needs, developed a prototype high-speed quantum random number generator small integrated. The developed quantum random number generator is generated when the speed reaches 3.2 Gbps, and by NIST random testing, the basic characteristics of true random number. In this paper, with a different random number generator are compared, finally chose the solution of quantum random number based on the phase fluctuation of laser, and the theoretical analysis on the feasibility of this scheme through modeling, and experimental methods to determine the optimal work point. Based on quantum random number of laser phase fluctuations occur scheme, quantum phase fluctuations as quantum signal source, the random laser from spontaneous emission. The conversion phase interferometer can not direct measurement for intensity fluctuation can be measured, and then through the high speed photodetector is converted to electrical signals and random fluctuations, were collected to get the quantized quantum random number primitive. In addition, this paper also designed for testing purpose memory module and optical fiber, Gigabit Ethernet, USB 2 data transmission interface to meet different application requirements, which through the effective output rate of fiber transmission channel quantum random number is 3.2 Gbps. in the development of high speed laser based on quantum phase fluctuations of the random number generator in the process, for the safety of information theory that postprocessing of original random number processing speed slow, presents a high-speed processing algorithm in the realization of FPGA, real-time postprocessing speed reached 3.36Gbps, solve the technical bottleneck by classical noise. The adverse effects of non ideal factors such as the sound and the actual device, true random number of original random number is often not ideal, need to go through the postprocessing to obtain quantum random number with true random number features. And the existing security information theory proved by postprocessing process leads to large amount of calculation, slow processing speed, so the original random number is generally only offline instead of high-speed real-time postprocessing, real-time speed which severely limits the quantum random number. To solve this problem, we calculated by carefully studying the characteristics of large Toeplitz matrix based on postprocessing, we propose a parallel algorithm based on matrix decomposition, significantly reduced the hardware logic the resources required by the proposed algorithm, successfully implemented in FPGA high speed large Toeplitz matrix based postprocessing operation, real-time postprocessing speed reached 3.36 Gbps. after the US Quantum random number processing of statistical homogeneity, self correlation, NIST test and random test, the basic feature of quantum random number with true random number test results show that, postprocessing algorithm is effective. The parallel implementation of high-speed FPGA postprocessing algorithm solves the postprocessing speed with the original random number generating speed seriously does not match the problem, make the quantum random number generator laser phase fluctuation can really produce quantum random number based on the high speed. The main innovation and innovative achievements are as follows: 1) to include the high quality quantum key distribution and other modern security communication protocol, science and technology fields, high speed random number generating demand and successfully developed a prototype high-speed quantum random number generator small integrated. The real-time speed reached 3.2 Gbps the development of quantum random number generator, and can Through NIST random testing, the basic characteristics of.2 with the true random number) according to the safety information theory that postprocessing of original random number problem of processing speed, the high speed processing algorithm based on matrix decomposition, the algorithm significantly reduces the logic resources required, successfully implemented in FPGA large Toeplitz matrix postprocessing operations, solve the high-speed random number postprocessing technology bottleneck in speed, and the formation of patent technology.

【学位授予单位】：中国科学技术大学
【学位级别】：博士
【学位授予年份】：2017
【分类号】：O413

【相似文献】