【作者】 邹艳；

【导师】 陆佩忠；

【作者基本信息】 复旦大学 ， 计算机应用技术， 2006， 博士

【摘要】 信息截获是电子信息学中十分特殊的技术领域，是第三方对通信双方的信息进行非常规获取的一种技术。有别于常规的通信技术，信息截获技术面临如下重大的基本难题：通信体制参数的盲识别；系统信号参数的盲识别；如何实现对强度低于常规通信接收门限的衰落信号的高增益；如何实现对混合信息流的高容错同步分路；以及如何实现对信息价值的判定。本文用代数理论和统计方法研究上述主要难题，给出恰当的模型描述和系统理论分析，和一系列独创性的快速算法，用于解决现实信息截获任务。具体地讲，本文在如下四个方面给出了含有众多重大创新性的深刻研究。1)在信道编码参数的盲识别研究方面取得了重要进展。如何仅从接收的数据流中反推出所采用的信道编码的编码体制，这是实现信息截获不可避免的难题。著名的Berlekamp-Massey（BM）算法实现对已知的一条线性递归序列快速地求出未知的生成多项式，这也是用于识别通信中常用的加扰方式的有效方法。本文构造了一个齐次关键模方程，实现对刻画BM算法有重要意义的关键方程作了全新的推广，并用有限域F上的两个变元的多项式环F[x，y]的齐次理想刻画该齐次关键模方程的解空间。从代数理论的角度上看，我们把BM算法推广成求代数模的极小生成元，给出了求Gr（?）bner基的快速算法，并由此给出了求解齐次关键模方程的快速算法。新算法具备如下的重要的创新功能和性质：（1）解决卷积码的快速盲识别问题；（2）计算复杂性是接收到序列长度的平方；（3）与BM算法不同，新算法具有一定的容错性，从而也转化成对盲识别计算效率的大幅度提升。2)给出多路信源的联合信源信道译码算法。信号的强纠错技术是信息截获的重要课题。对没有纠错编码设计或含有信道编码设计但纠错能力不够的微弱信号的截获，如何提升信号的增益，需要探索有别于传统的信道译码和信源译码相分离的译码新方法。本文研究非对称的信源信道联合译码方法，提出了一种多路可变长编码信源复接情况下的联合译码方法，合理利用多路信源的先验信息来增强信道译码性能。目前针对单一信源的联合译码的研究较多，而多路信源复接是现代宽带无线通信中非常常用的技术，因此本文提出的针对多信源的联合译码方法更贴近实际应用。本方法原理上推广了Viterbi译码算法，形成一个需要兼顾信道编码序列流的最优路径与多路信源编码状态转移树中的最大先验概率的译码算法。实验结果表明，本方法比不利用先验信息的普通Viterbi译码算法有大约1.5dB的性能提升。3)解决了高误码率环境下的扩频码的捕获问题。对扩频信号进行信息截获首先需要解决扩频码的同步问题。本文提出了软判决的快速相关攻击新算法，该方法是密码分析领域的硬判决相关攻击算法在通信领域的巧妙推广与应用。利用软判决的快速相关攻击新算法，本文解决了在低信噪比环境下扩频通信中扩频码的快速捕获问题。实验结果表明，该算法能在极低信噪比情况下以很少的数据量快速准确的实现同步捕获。另外我们还将软判决快速相关攻击算法应用于抗几何攻击的图象隐秘通信，使得隐秘图象的抗几何攻击能力也有明显提升。4)相当彻底地解决高容错的数字分接问题。对数字群落信号进行信息截获必须事先确保信息帧结构的同步，正确的数字分接是信息还原的前提。本文研究了信息截获背景下的高容错的数字分接技术，提出了基于统计的帧同步定位方法和基于有限状态自动机的最大后验概率码速调整预测算法。实验结果表明，在误码率为2×10^-2时仍能实现稳健的分接，而在误码率高达5×10^-2时同步维持时间也能达到15分钟。而国际标准推荐的方法，在P_e=2×10^-3时，同步维持时间为大约1分钟。而且本方法所需要的计算量很小，因此本文给出的新方法可以实现高容错高速的信息截获。更多还原

【Abstract】 Information intercepting is a very special field in electronic informatics, which is an unconventional technique to acquire information from the communication between the transmitter and the receiver as a third party. Different to the conventional information communication, Information intercepting technique have to be faced with the following basic difficult problems: the blind recognition of the parameters of the communication; the blind recognition of the parameters of the signal; how to efficiently process the fading signal which power is under the receiving threshold of the conventional communication; how to demultiplex the mixed information stream with serious errors; and how to determine the value of the information etc. In this thesis we study the main problems above using algebraic theory and statistical method, we present proper modeling description and system theoretic analysis as well as a series innovative fast algorithms to solve the practical problem in information intercepting. In the concrete, we give an in-depth study with some great innovations in the following four aspects.1) Blind recognition of the parameters of the channel coding.How to reversely deduce the channel code in used only from the receiving data stream is an unavoidable difficulty in information intercepting. The famous Berlekamp-Massey （BM） algorithm can fast compute the unknown generator polynomial from a linear recursive sequence （LRS）, which is an efficient method to recognize the regular scrambling type in communication. We construct a HomogenousKey Module Equation （HKME） which is described by homogenous ideal of F[x,y],where F is a filed. HKME is a new generalization of the Key Equation （KE） that is very important to depict the BM algorithm. From the aspect of algebraic theory, we generalize the BM algorithm to the problem of computing minimal generators of algebraic module and present a fast algorithm of computing the Grobner basis, then give a fast algorithm of solving the HKME. The new algorithm has some important innovations and properties as following: （1） can solve the problem of fast blindrecognition of convolutional codes; （2） the computational complexity is O（N²）,where N is the length of receiving sequence; （3） different to BM algorithm, the new algorithm is error-resilient in a certain extent, then the efficiency of blind recognition is greatly improved.2) Joint source-channel decoding （JSCD） of multiple sources.Strong error-correcting technique is an important issue in information intercepting. To acquire the fading signal which is not designed with error-correcting code or the designed channel coding is not strong enough and improve the gain of signal, we need to investigate a new decoding method different to the current scheme that source decoding and channel decoding is separate. We focus on the unsymmetrical JSCD and present a method on joint decoding of multiplevariable-length encoded sources. By exploiting the a priori redundancy in multiple sources, the decoding performance is greatly improved. Currently most researches on JSCD are concern with single source. However the multiplex of multiple sources is a commonly used technique in modern wide-band wireless communication. Therefore the proposed method is more practical. Our method generalize the conventional Viterbi decoding algorithm, not only the optimal path in channel decoding trellis but also the a priori probability of branches in multiple VLC source trees are considered. The Simulation results show that our new method obtains the gain in SNR of around 1.5dB comparing to the conventional Viterbi algorithm that not exploits the a priori information.3) Fast acquisition of the spread-spectrum pseudonoise （PN） code from the highly noisy channel.Information intercepting from the spread-spectrum signal needs to synchronize the PN code first. We present a novel soft fast correlation attacks （SFCA） algorithm, which is a good generalization and application of hard fast correlation attacks in cryptanalysis field to communication field. With the novel SFCA algorithm, we solve the problem of fast acquisition of PN code in spread-spectrum communication in very low SNR circumstance. The simulation results show that the proposed SFCA algorithm can accomplish the fast acquisition correctly with few chips in very low SNR. In addition, we apply the SFCA to the image data hiding scheme against geometrical distortion, which enhance the ability of resisting geometrical distortion obviously.4) Robust error-resilient demultiplexing of digital signal.Information intercepting from the digital group signal needs to synchronize the information frames and correct demultiplexing is the precondition of information recovery. We study the robust error-resilient digital demultiplexing techniques for information intercepting. We present a frame alignment method based on statistic and a maximum a posteriori （MAP） method for justification prediction based on finite state automaton. The simulation results show that our methods can realize robust demultiplexing with the bit error rate of 2 × 10^-2 and the synchronization maintaining time can reach 15 minutes when the bit error rate up to 5× 10^-2. However with the standard method recommend by ITU, the average time between losses of synchronization is about 1 minute. The computational complexity of our method is very little. Therefore our method can accomplish robust and fast error-resilient information intercepting.更多还原