【作者】 别志松；

【导师】 吴伟陵；

【作者基本信息】 北京邮电大学 ， 信号与信息处理， 2007， 博士

【摘要】 迭代接收技术能够大大提高系统性能，因子图是一种用于分析和设计迭代接收机的有效工具。本论文对于两种基于因子图的迭代接收结构进行了深入的研究。主要工作包括：1．对IDMA原理进行深入研究，包括：(1)运用因子图对交织分多址接入(IDMA)的迭代多用户检测问题进行分析，并与CDMA的因子图相比较，分析IDMA能够采用低复杂度迭代检测算法的原因，(2)用和积算法的一般准则推导IDMA检测算法，揭示IDMA低复杂度检测算法与因子图上和积算法的内在联系。(3)用EXIT图对未编码IDMA系统和卷积编码IDMA系统进行分析，研究信道编码方式对于IDMA系统性能的影响。2．对等功率IDMA系统进行了优化设计，包括：(1)对未编码IDMA系统中消息之间的相关性进行分析，在此基础上设计了一种PEG交织器，改进未编码IDMA系统性能；(2)将LDPC码用于多址接入信道，设计了多用户LDPC编码和LDPC编码IDMA两种方案，设计了一种针对IDMA低复杂度检测算法的高斯近似密度进化方法，将这种方法与差分进化方法相结合对用于多址接入信道的非规则LDPC码进行优化。3．设计了一种用于B3G系统的基于IDMA的单步随机接入信道。将IDMA用于B3G系统随机接入信道的资源请求数据传输。设计了完整的单步随机接入信道，包括前导码和资源请求数据，提出了一种基于峰均比检测的前导码检测算法并对其性能进行分析；通过在数据部分加入子标识的方法，使得基于IDMA的单步随机接入具有更低的碰撞概率。4．研究了m序列迭代捕获有关的三个问题，包括：(1)非相干差分迭代捕获算法及序列还原：利用m序列的差分特性，设计了一种差分和积迭代捕获算法，并利用m序列的相差约束特性，设计了一种低复杂度的差分序列还原算法。(2)多个m序列的并行迭代捕获问题：将m序列的迭代捕获和LDPC多用户编码结合起来，设计了一种低复杂度的多个m序列并行迭代捕获算法。(3)生成多项式非“稀疏”的m序列并行迭代捕获的因子图优化问题：利用m序列的采样特性和相差约束特性，结合LDPC码因子图的优化设计理论，对生成多项式非“稀疏”的m序列因子图进行优化设计，获得了良好的迭代捕获性能。更多还原

【Abstract】 Iterative receiver technology can improve the system performance dramatically. Factor graph is a powful tool to analyze and design iterative receiver. In this thesis, several iterative systems are analyzed intensively. The main topics include:1．IDMA systems are studied intensively. The main contributions include: (l)The problem of iterative multiuser detection for IDMA system is analyzed by using factor graph. The factor graph representation of IDMA system is compared with that of CDMA system. The reason for that IDMA can use low complexity detection algorithm is analyzed. (2) The detection algorithm of IDMA system is derived by using the universal rule of sum-product algorithm and the inner relationship between the IDMA detection algorithm and the sum-product on factor graph is also discovered. (3) EXIT charts technology is used to analyze the performance of uncoded IDMA and convolutional coded IDMA systems. The effect of channel coding scheme on IDMA systems is studied.2．IDMA system with equal power constrains is optimized. The main contributions include: (1) Correlation among the messeges in uncoded IDMA systems is analyzed. A kind of interleavers, named by PEG interleavers is designed for uncoded IDMA system. (2) LDPC codes are used to multiple access channel. Two schemes, named by multiuser LDPC coding and LDPC-coded IDMA, respectively, are proposed and the corresponding message update rules for three kinds of nodes are designed. A Gaussian approximation method is designed these update rule. Differential evolution, combined with Gaussian approximation technology, is used to optimize the irregular LDPC codes for these schemes.3．An IDMA based one step random access channel structure is designed for B3G systems. IDMA is used as the transportation mode for the resource request message part. The whole RACH channel structure, including preamble part and message part, is designed for one-step random access. The preamble detection algorithm is designed and analyzed. By inserting several bits in the message part as addition ID, the IDMA-based one step access scheme has lower collision probility than conventional schemes.4．Three problems about iterative acquisition for m sequence are studied. The main contributions include: (1) Non-coherent differential iterative acquisition for m sequence. By using the differential property of m sequence, a differential iterative sum-product acquisition algorithm is designed. A simple method to find original sequence phase with known differential sequence phase is designed. (2) The problem of parallelized iterative acquisition for multiple access. By combining the low complexity multiuser detection algorithm for IDMA and the sum-product iterative acquisition algorithm for m sequences, a parallelized iterative acquisition algorithm is proposed for multiple m sequences. (3) The problem of iterative acquisition for m sequence with non-sparse generating polynomial. By using the sampling property and the phase difference constrains property of m sequenece, aided by the LDPC codes optimization theory, the factor graph representation of m sequenece with "non-sparse" generating polynomial is optimized and good acquisition performance is acquired.更多还原