【作者】 林毅；

【导师】 曾贵华；

【作者基本信息】 上海交通大学 ， 通信与信息系统， 2013， 硕士

【摘要】 在当今的信息社会中，通信技术发挥着越来越重要的作用，同时人们对通信安全性也提出了越来越高的要求。经典密码学是保障信息安全的有效工具，然而随着计算机和量子计算的发展，基于数学计算复杂性假设的经典密码体制日益受到严峻的挑战。量子密码学建立在量子力学原理基础上，被证明能够提供信息论意义上的绝对安全性。量子密钥分发(QKD)作为量子密码学的一种重要应用，在量子测不准原理和不可克隆性定理保障下，使合法通信双方Alice和Bob能够在存在窃听者Eve的情况下建立无条件安全的共享密钥。QKD包括量子信道传输、数据筛选、密钥协商和保密增强等步骤，其中密钥协商和保密增强合称为后处理。后处理算法对QKD的密钥速率和安全距离起着至关重要的作用。连续变量QKD相比离散变量QKD具有设备更简单，信道容量更高等优点，具有更好的应用前景，但连续变量QKD优越性的体现需要依赖更为复杂的连续变量后处理算法的性能。本文致力于连续变量QKD后处理算法，特别是密钥协商算法的研究。通过研究常见后处理算法的具体实现细节，提出改进方案，提升算法性能，并对算法进行软件实现。主要工作如下：1.对Slice协商算法的整个过程，包括量化区间划分、估计函数设计、效率计算等方面进行了详细的分析和讨论，对量化区间划分算法进行了改进。将层间迭代算法应用到Slice协商框架中，并针对LDPC码设计了有效的译码初始概率计算方法，提高了协商效率。此外，还对Slice协商框架中使用Cascade算法的方案进行了仿真研究，验证其可行性。2.对LDPC码在多维协商算法框架中的应用方案进行了研究，分析了等效信道噪声模型，针对二进制LDPC码和多进制LDPC码分别设计了适用的后验概率计算方法，并通过仿真验证了多进制LDPC码能提供更好的协商性能。3.设计并实现了一个后处理算法软件，该软件能够实现基本的密钥协商和保密增强功能，并具有良好的扩展性，为后续开发高效后处理软件打下基础。更多还原

【Abstract】 In today’s information society, communication technology isbecoming more and more important, and the demand for informationsecurity becomes higher. Classical cryptography is useful to protectinformation, whereas it faces challenges from the development ofcomputer technology and quantum computation, due to its hypothesis oncomputationally complexity. By contrast, quantum cryptography is built onthe basic principles of quantum mechanics, thus providinginformation-theoretic security.Quantum key distribution (QKD) is an important application ofquantum cryptography, which enables two legal parties，Alice and Bob，toshare a absolutely secure key, in spite of the existence of a eavesdropperEve，guaranteed by the uncertainty principle and quantum no-cloningprinciple. A QKD protocol includes four main steps，that is，quantumchannel transmission，data sifting，reconciliation and privacy amplification.The last two steps are generally together referred to as postprocessing andits performance greatly influences the key rate and secure communicationdistance of QKD.Continuous variable QKD has advantage of simpler equipment,higher achievable channel capacity and so on, thus has a better applicationprospect compared to discrete variable QKD. However, the realizationrequires more efficient continuous variable postprocessing algorithmswhich are much more complex.In this thesis, we focus on the research of postprocessing algorithm inthe continuous variable QKD, especially the reconciliation algorithm. We investigate details of several postprocessing algorithms,try to make someimprovements, and implement the algorithm using software. The mainwork is as follows.1.We deeply analyse and discuss the process of Slice reconciliation,including the partition of interval, design of estimating function, andcalculation of reconciliation efficiency. An improvement on intervalpartition is presented. After that, we apply inter-layer iteration to the Slicereconciliation, and design an effective way for the computation of initialprobability used in the decoding of LDPC codes. Also, we investigate thescheme of embedding Cascade algorithm in Slice reconciliation throughsimulation, and verify its performance.2.We study the application of LDPC codes in the multidimensionalreconciliation algorithm，analyse the noise model of equivalent channel，and design practical formula to calculate the posterior probability fordecoding of binary and non-binary LDPC codes. Through simulation, wedemonstrate that a performance gain can be achieved by using non-binarycodes.3.We design and implement a postprocessing software. This softwarecan realize the fuction of reconciliation and privacy amplification. Also,due to its good extensibility,future development of postprocessing softwarewith higher performance can be built on its framework.更多还原