【作者】 王正海；

【导师】 田茂；

【作者基本信息】 武汉大学 ， 通信与信息系统， 2011， 博士

【摘要】 在无线通信中,Turbo码是目前应用最广泛的信道编码技术之一,它明显提升了无线频谱的使用效率。分析发现,交织增益的存在使得Turbo码的纠错性能依赖于其采用的交织器的交织深度。以第四代宽带无线通信系统LTE-A为例,在码率R=1/3、QPSK调制及误块率为10-4的条件下,和较短(K≤200,K是交织器的交织深度)的Turbo码相比,较长(K≥1000)的Turbo码能获得1.5-3dB的增益。为了改进短Turbo码纠错性能,本文研究了Turbo码优化的网格终止方法、短Turbo码内部交织器设计方法、多用户协作Turbo编码方法和多用户协作自动请求重传方法。本文的具体贡献如下：1)双结尾Turbo码是目前最常用的Turbo编码方法,其分量编码器的网格终止方法会导致边界效应以及额外的尾比特开销。这些缺点影响了Turbo码的性能,尤其是交织深度较短和码率较高的Turbo码的性能。本文通过Turbo码分量编码器之间的连续编码,达到了降低边界效应和尾比特开销的目的。理论分析和仿真验证表明：Turbo码分量编码器之间的连续编码能够优化Turbo码网格终止的性能,进而提高Turbo码的性能,特别是提高交织深度较短和码率较高Turbo码的性能。2)对于交织深度较短的Turbo码,其分量编码器产生的低重量校验比特序列是由具有长度小于(?)的自终止输入模式的输入信息比特序列导致的。本文通过对Turbo码自终止输入模式的分析,发现最大扩展因子互素交织器能够有效地打破所有长度小于(?)的自终止输入模式。理论分析和仿真验证表明：采用最大扩展因子互素交织器作为Turbo码的内部交织器能够提高交织深度较短的Turbo码的性能。3)相对于单用户处理技术,多用户联合处理使得参与联合处理的每个用户能够共同利用多个用户的资源以协作的方式来获得更大的增益,本文提出了一种多用户协作Turbo编码方法。在采用多用户协作Turbo编码的下行链路中,在给定的服务可靠性要求下,接收端可靠译码需要的对数信噪比与参与协作的用户数N和消息长度K的乘积成反比。即：参与协作的用户数N和消息长度K的乘积越大,接收端可靠译码需要的对数信噪比值就越低。另外,参与协作的用户数N和消息长度K的乘积越大,性能增益的增量就越小。理论分析和仿真验证表明：本文给出的多用户协作Turbo编码方法适合在用户的消息长度K较短的场景中使用。4)基于无线电波的广播特性和网络编码,本文研究了一种新的多用户自动请求重传方法。基站将多个用户的重传数据包逐比特地线性组合在一起后再广播；用户利用置信传播算法从组合数据包中检测出基站发给本用户的重传数据包,并和以前传输的数据包软合并后再译码。理论分析和仿真证明：在复杂度和单用户自动请求重传相近的情况下,多用户自动请求重传能够将无线通信系统的频谱效率提高8%。采用和广播链路上多用户自动请求重传类似的原理,本文还讨论了单播链路上多个并行数据流的联合自动请求重传问题。在多数据流自动请求重传中,发射端采用网络编码将多个数据流的重传数据包经过合并后再传输给接收端。基于置信传播算法,接收端从组合数据流中检测出各个数据流的重传数据包,并和在前次传输时缓存的数据流软合并后译码。此外,接收端再用置信传播算法将先译码成功的数据流从组合数据流中消除,以提高未成功译码数据流的检测性能。和单播链路上传统的多个数据流独立自动请求重传相比,多数据流联合自动请求重传能够将单播链路的频谱效率提高10%,而仅增加不到20%的复杂度。综上所述,本文以提升Turbo编码无线通信系统的频谱效率为目标,优化了双结尾Turbo码的网格终止方法,研究了短Turbo码内部交织器的设计方法,给出了一种多用户协作Turbo编码方法和多用户协作自动请求重传方法。本文的工作使得Turbo编码无线通信系统能够更好地支持各种长度的业务数据包,具有重要的学术意义和应用价值。尤其是,本文探讨的短Turbo码内部交织器的设计方法和多用户协作Turbo编码方法能够明显优化短Turbo码的性能,显著提升Turbo编码无线通信系统对短数据包的传输能力。更多还原

【Abstract】 Turbo codes, which are among the most widely used channel coding techniques, have significantly improved the spectrum efficiency of many wireless communication systems. Extensive studies have shown that the interleaving performance gain makes the error correction capability of turbo codes be relevant to the depth of the turbo codes inner interleaver. E.g., assuming the code rate, modulation type, and block error rate to be R=1/3, QPSK, and 10-4, respectively, the LTE-A turbo codes using the interleavers with large depth (i.e., K≥1000, where K is depth of the turbo codes inner interleaver) would outperform the ones using the interleavers with short depth (i.e., K≤200,) by1.5-3dB. For the sake of simplicity, we name the turbo codes using the interleavers with short depth by short turbo codes in this thesis, otherwise large turbo codes for short. To enhance the error correction ability of short turbo codes, we have studied the optimization of trellis termination for turbo codes, design of interleavers for short turbo codes, multi-user cooperative turbo coding, and multi-user ARQ. The main contributions are:1) Dual terminated turbo codes are the most popular turbo coding techniques. However, the trellis termination of dual terminated turbo codes would incur edge effect and extra tail bits, which will negatively impact the performance of turbo codes, especially the performance of short and high rate turbo codes. Hence, the continuous coding between both component encoders is used to mitigate the edge effect and decrease the length of tail bits, and two decoding algorithms are applied to decode the new coding scheme. Analyses show that the continuous coding between the both component encoders of dual terminated turbo codes can optimize the trellis termination of turbo codes, and finally improve their performance, especially the performance of short and high rate turbo codes.2) The low weight parity sequences of both component encoders of turbo codes are produced by the input sequence with self-terminated patterns of length less than-(?)k/2. Analyses of input sequences with self-terminated pattern show that maximum spread factor relative prime interleaver (MSF-RP) can guarantee to break down all input sequences with the self-terminated pattern of length less than ()k/2. Hence, MSF-RP interleaver can enhance the error correction abilities of short turbo codes.3) Compared to the techniques based on the orthogonal temporal-spectral-spatial resource assigned for every single user, more gain can be achieved by multi-user cooperation, which enables every one in the cooperation to explore more resource in a cooperative manner. Herein, we have investigated the multi-user cooperative turbo coding. Analyses show that, given the quality of service (QoS) in downlink, the log-signal-to-noise-ratio (SNR) is inversely proportional to the product of N and K, where N and K denote the number of users in cooperation and the message length of every user, respectively. That is, the larger the product of N and K, the lower the required log-SNR. Beside, the larger the product of N and K, the less the gain of the improvement. Hence, multi-user cooperative turbo coding is suitable for the scenario where the length of the message is relatively short.4) Based on the broadcasting nature and network coding, we have studied multi-user ARQ in downlink, where the base station broadcasts the linear combination of the retransmission packets from multi-user, and the retransmission packet is detected by belief propagation (BP) algorithm and softly combined with the previously received packet at the intended users before decoding. Analyses show that compared to single user ARQ, the spectrum efficiency can be improved by 8% by multi-user ARQ; meanwhile, the complexity increasing is negligible. Using the similar principle, we have extended the multi-user ARQ on the broadcasting link to the unicast link, where the sender transmits the linear combination of the retransmission packets from multi-stream by network coding, and the retransmission packet is detected by BP algorithm and softly combined with the previously received packet of the intended stream at the receiver before decoding. Furthermore, the retransmission packets of correctly recovered streams are canceled from the composite packet to improve the further decoding performance of the residual retransmission streams. Analyses show that, compared to single stream ARQ on the unicast link, the spectrum efficiency can be improved by 10% by multi-stream ARQ, meanwhile, the complexity increasing is less than 20%.In summary, aimed at improving the spectrum efficiency of turbo coded wireless communication systems, we have optimized the trellis termination for dual terminated turbo codes, designed the inner interleavers for short turbo codes, investigated the multi-user cooperative turbo coding, and studied multi-user ARQ. The techniques proposed in this thesis can optimize the performance of turbo coded wireless communication systems need to transmit traffic packages of various size, and have important practical and academic significance. Particularly, the performance of short turbo codes can be obviously improved by the contributions in this thesis, e.g., short turbo codes with MSF-RP interleavers and multi-user cooperative turbo coding. Hence, our work can significantly promote the abilities of turbo coded wireless communication systems to transmit short traffic packages.更多还原