量子密钥分发后处理关键技术研究

发布时间：2017-12-28 00:20

本文关键词：量子密钥分发后处理关键技术研究　出处：《哈尔滨工业大学》2016年博士论文　论文类型：学位论文

【摘要】：密码技术是保障信息安全的重要手段,但是现有的密码技术通常是计算安全的。目前唯一被证明是无条件安全的密码算法是一次一密(One Time Pad,OTP),但是在实际应用中OTP存在密钥分发难的问题。量子密钥分发(Quantum Key Distribution,QKD)利用量子力学原理成功解决了密钥分发难的问题,使OTP算法有了实际的用武之地。QKD由量子部分与经典后处理部分组成。量子部分以量子态为载体在通信双方之间分发部分安全、部分相关的原始码,经典后处理对这些原始码在认证的经典信道上进行筛选、误码协商、私密放大处理,最终得到无条件安全的密钥。QKD技术的目的是为通信参与者提供无条件安全的密钥分发,因此提高净安全密钥分发速率是其首要任务。作为QKD的关键组成部分,QKD后处理理所当然要服务于这一首要任务。QKD系统由于通信距离、协议设计等因素的不同,输入到经典后处理的原始码的特点也不尽相同。因此在给定原始码序列时,如何通过优化QKD后处理各模块以使净安全密钥分发速率最大化是一个亟需解决的问题。本文的主要研究内容和创新性如下:(1)以净安全码率最大化为目标,通过逐模块地分析后处理对原始码数据的处理过程及其对净安全码率的影响,建立了QKD后处理性能优化模型。基于该模型,我们为后处理各模块分别提出了评价指标。在提出的评价指标下,我们对目前的研究热点——误码协商算法进行了性能评估。此外,在该性能优化模型的指导下,我们对后续的三个研究内容进行了研究。(2)针对高重复频率QKD系统的筛选模块面临处理数据量大、认证密钥消耗量高的问题,提出了一种基于MZRLFL编码的高压缩比的筛选算法。提出的算法从分析筛选模块通信数据的特点出发,充分考虑了筛选模块在处理速率与存储方面的压力,具有接近香农限的压缩性能。MZRLFL编码首先利用修改的游程长度编码将二元信源转换为n元信源,再利用可高速编解码的定长编码对n元信源的消息进行编码。为了弥补定长编码压缩性能的不足,我们通过寻找最优参数使其压缩性能达到最佳。理论分析与实验结果表明,提出的算法具有接近香农限的性能。为了验证压缩性能对认证密钥消耗的影响,将提出的算法应用于实际QKD系统。分析表明,当通信距离为1km与25km时,后处理认证密钥消耗量分别减少了26%与15%以上。(3)针对高筛选码速率的QKD系统需要高速的误码协商算法,研究了基于极化码的误码协商算法。我们分析了极化码在QKD中应用模式的安全性及性能,分析表明,极化码最优的两种应用模式的延迟只与译码操作有关。因此我们接着对SC译码算法进行了三方面的优化以提高译码速率,其中前两个优化主要适于面向硬件实现的SC译码算法,第三个优化主要适于面向软件实现的SC译码算法。为了给优化提供理论支撑,我们分析了SC译码算法中似然比计算的依赖关系。提出的三方面优化为:(i)针对现有的SC译码调度算法仍很原始这一问题,提出了一种高效的SC译码调度算法,与现有算法相比,该算法具有关键路径延迟与空间复杂度均为常数等优点,能够降低硬件实现的存储资源消耗,并提高硬件实现的时钟频率,从而提高译码速率;(ii)针对SC译码算法译码延迟高的问题,基于预先计算的思想提出了一种低延迟的SC译码算法,并详细分析了其代价,理论分析表明该算法可以以很低代价大大降低译码延迟;(iii)针对面向软件实现的SC译码算法均使用递归的方式这一问题,提出了一种非递归的SC译码算法,实验结果表明非递归的SC译码算法的译码速度是递归的SC译码算法的2.2倍至3.3倍。(4)针对低筛选码速率的QKD系统需要高协商效率的误码协商算法,提出了一种高协商效率的误码协商算法。鉴于在已有的误码协商算法中,Cascade类算法在协商效率方面通常具有最好的性能。秉着尽可能利用已有信息推断分组的奇偶性,避免无谓的信息泄露的思想,我们从两方面优化了Cascade类算法。首先,我们证明了Cascade类算法利用回溯纠错技术对每个分组进行纠错时总是纠正偶数个错误,利用该性质我们可以推断第i轮(i大于等于2)的最后一个分组的奇偶校验码总是相同,无需进行交互比对。其次,我们发现,在Cascade类算法回溯纠错时,对于长度为2的分组可以不执行BINARY过程而直接完成纠错。这两方面的优化减少了Cascade类算法纠错过程中的信息泄露量,提高了协商效率。
[Abstract]:Cryptography is an important means to ensure information security, but the existing cryptography is usually secure. At present, the only proven cryptographic algorithm is unconditionally secure is a secret (One Time, Pad, OTP), but in the actual application of OTP are key points of attack. Quantum key distribution (Quantum Key, Distribution, QKD) by using the principle of quantum mechanics has successfully solved the key problems in the OTP algorithm, with the actual use. The QKD is composed of the quantum part and the classical post-processing section. The quantum part distributes some secure and partially related codes between the two sides with the quantum state as the carrier. After classical post-processing, the original codes are screened, the error codes are negotiated, and the private amplification processes are processed. Finally, the unconditionally secure key is obtained. The purpose of QKD technology is to provide unconditionally secure key distribution for communication participants, so improving the net secure key distribution rate is its primary task. As a key component of QKD, QKD post processing is of course to serve this primary task. Because of the difference of communication distance, protocol design and other factors, the characteristics of the original code input to the classic post processing in QKD system are different. Therefore, it is an urgent problem how to maximize the net security key distribution rate by optimizing the QKD after the optimization of the original code sequence. The main contents and innovations of this paper are as follows: (1) with the goal of maximizing net security rate, we analyze the process of post processing on the raw code data and its influence on net security bit rate by modular analysis, and establish a QKD post-processing performance optimization model. Based on this model, we put forward evaluation indexes for each module respectively. Under the proposed evaluation index, we evaluate the performance of the current research focus, the error code negotiation algorithm. In addition, under the guidance of the performance optimization model, we have studied the following three research contents. (2) aiming at the problem of high data volume and high consumption of authentication key for screening module of high repetition rate QKD system, a new algorithm based on MZRLFL coding for high compression ratio is proposed. The proposed algorithm takes account of the characteristics of the communication data of the screening module, and fully considers the pressure of the screening module in terms of processing speed and storage, and has the compression performance approaching the Shannon limit. MZRLFL coding first uses the modified run length encoding to transform the two element source into the N source, then encodes the message of N meta source by using the fixed length encoding with high speed encoding and decoding. In order to make up for the shortage of the compression performance of the fixed length coding, we make the best compression performance by finding the optimal parameters. Theoretical analysis and experimental results show that the proposed algorithm has the performance of close to the Shannon limit. In order to verify the effect of compression performance on authentication key consumption, the proposed algorithm is applied to the actual QKD system. The analysis shows that when the communication distance is 1km and 25km, the post processing authentication key consumption is reduced by 26% and more than 15%, respectively. (3) for QKD system with high filter rate, a high speed error negotiation algorithm is needed, and the error code negotiation algorithm based on polarization code is studied. We analyze the security and performance of the application mode of polarization code in QKD. The analysis shows that the delay of the two best application modes is only related to the decoding operation. Therefore, we have optimized the SC decoding algorithm in three ways to improve the decoding rate. The first two optimizations are mainly suitable for the SC decoding algorithm for hardware implementation, and the third optimization is mainly suitable for the SC decoding algorithm for software implementation. In order to provide theoretical support for optimization, we analyze the dependence of the likelihood ratio calculation in the SC decoding algorithm. Three optimization is put forward: (I) SC for decoding the existing scheduling algorithm is still very primitive this problem, proposes a SC decoding efficient scheduling algorithm, compared with the existing algorithms, this algorithm has the critical path delay and space complexity are constants and other advantages, can reduce storage resources low hardware implementation consumption, and improve the clock frequency of the hardware implementation, so as to improve the decoding rate; (II) for SC decoding algorithm, the decoding delay high, pre computation proposed a low delay based on the SC algorithm, and a detailed analysis of the price, theoretical analysis shows that this algorithm can greatly reduce the cost of a very low the decoding delay; (III) the problem for SC decoding algorithm oriented software using recursive way, proposes a non recursive SC algorithm, the experimental results show that the decoding speed of decoding SC non recursive algorithm is recursive 2.2 to 3.3 times the SC decoding algorithm returned. (4) a high negotiation efficiency error negotiation algorithm is proposed for QKD system with low filter rate, which requires high negotiation efficiency. In view of the existing error negotiation algorithms, the Cascade class algorithm usually has the best performance in terms of negotiation efficiency. We optimize the Cascade class algorithm from two aspects, in order to use the existing information as far as possible to deduce the parity of the packet, and to avoid the meaningless information disclosure. First, we show that the Cascade algorithm with backtracking always correct even a mistake correcting technique for error correction for each packet, we can infer the properties of the I wheel (I is greater than or equal to 2) parity check code of the last packet is always the same, without the need for mutual comparison. Secondly, we found that in the Cascade class algorithm backtracking error correction, for the length of 2 of the packet can not perform the BINARY process and directly complete the error correction. The two aspects of the optimization reduce the information leakage in the Cascade class algorithm, and improve the negotiation efficiency.
【学位授予单位】：哈尔滨工业大学
【学位级别】：博士
【学位授予年份】：2016
【分类号】：TN918.4;O413

【参考文献】