【作者】 李强；

【导师】 李少谦；

【作者基本信息】 电子科技大学 ， 通信与信息系统， 2008， 博士

【摘要】 采用迭代译码技术是Turbo码和低密度校验码能获得接近香农限性能的主要原因之一，采用高效纠错码和迭代检测技术是实现高速数据传输、提高频谱利用率和对抗各种干扰与衰落的一种有效途径，是未来新一代宽带移动通信系统的主要候选技术之一。本文首先研究了迭代译码与迭代检测技术的发展与现状，分析了目前该领域需要进一步解决的一些关键问题，列出了本文的研究内容。本文的第二部分研究了信道信噪比估计误差对时域迭代均衡系统性能的影响。通过分析最优的对数后验概率迭代均衡算法和次优迭代均衡算法输出的软信息，本文提出了一种缩放软信息Max-Log-MAP迭代均衡算法，该算法性能在低损失信道上逼近最优算法，同时新算法对信噪比误差不敏感。针对高损失信道，本文又提出了一种基于软输出信息的信噪比估计算法。该算法利用了次优迭代均衡算法对信道信噪比误差不敏感特性，从其软输出信息中提取信道信噪比。利用上述信噪比估计算法，本文提出了一种切换迭代均衡算法，该算法性能逼近已知信道信噪比的优化算法。本文第三部分研究了频域迭代均衡算法，提出了一种新的并行级联频域均衡结构。基于最小均方误差判决反馈均衡准则，提出了该并行级联结构的硬判决反馈和软判决反馈迭代均衡算法，分析了该算法在静态信道和瑞利衰落信道上的性能，结果表明，本文提出的迭代频域均衡算法比传统的迭代频域判决反馈算法较大的性能改善。此外，本章还分析了信道估计误差和信噪比不匹配对新算法的影响。本文第四部分研究了迭代检测技术在军事通信中的抗人为干扰的性能。针对慢速跳频连续相位调制系统，本文提出了一种对信道信噪比误差不敏感的软输出软输出解调算法，通过校正译码器输出软信息，基于新解调算法的迭代解调与译码算法性能非常接近最优的迭代检测算法，而且新算法不受信噪比不匹配的影响。特别地，针对部分频带干扰问题，本文又提出了一种迭代干扰估计算法，分析了迭代解调和译码算法结合干扰估计算法的抗干扰能力。仿真结果显示了迭代干扰估计算法大大改善了迭代解调和译码的抗干扰性能。最后，本文研究了B3G/4G系统中LDPC的迭代译码和设计中的关键问题。首先本文提出了一种半随机LDPC码构造方法，这种方法构造出的LDPC码具有准循环结构，同时参数选择比传统方法更为灵活；其次，本文研究了LDPC码的定点实现问题，分析了定点误差对译码算法性能的影响；利用π旋转LDPC码的特点，本文还提出了一种自适应LDPC编码策略，研究了它在衰落信道上的性能；最后，本文针对B3G TDD下行链路，提出了一种迭代多入多出检测和LDPC译码算法，研究了它在COST207信道上的性能。更多还原

【Abstract】 Iterative decoding algorithm has been introduced in advanced forward-error-correct-ing codes,which is just the reason why Turbo and LDPC codes have performanceclose to Shannon limits.Turbo principle can be used not only to offer the potential ofhigh spectral efficiency,but also to decrease the effect of fading and jamming.Therefore,iterative detection algorthim is one of the key candidate techniques of thenext generation broadband mobile communication system.Firstly,the researches of iterative decoding and detection are surveyed,a variety offundamental problems needed to be solved are investigated,and the main contents ofthis dissertation are list.In the area of Turbo equalization,the effect of SNR mismatch on bit error rateperformance is investigated.Through the comparative study between log maximum aposteriori probability (Log-MAP) and suboptimal Max-Log-MAP algorithm,a newiterative equalization algorithm called scaled Max-Log-MAP algorithm is proposed,which not only performs almost the same as that of Log-MAP over low-loss channels,but also is insensitive to a SNR estimation error.Over high-loss channels,a new SNRestimation method is proposed based on the reliability information of soft output fromscaled Max-Log-MAP equalizer.By using the estimated SNR,a new switchingiterative equalization algorithm between Max-Log-MAP and Log-MAP is alsoproposed.In the area of frequency-domain equalization,a novel parallel concatenated frequ-ency-domain equalization structure is proposed.By exploiting minimum mean squareerror decision-feedback equalization criterion,a new iterative frequency-domainequalization algorithm with hard-decision feedback and soft-decision feedback isderived,respectively.The performance of our new scheme is also investigated overboth static channels and Rayleigh channels.Moreover,the influence of a SNRmismatch on this new scheme is also analyzed.In the area of resisting man-made jamming,a slow frequency-hopping CPM system is considered.With the MAP demodulation algorithm,a new branch metricscalculation method which makes it insensitive to the SNR offset is presented.Thisalgorithm is also improved so that it can perform close to optimum MAP demodulationalgorithm.In aspect of anti-jamming,an iterative jamming estimation method isderived.The performance of the iterative CPM demodulation and decoding algorithmwith jamming estimation algorithm is analyzed.Simulation results demonstrate that theproposed jamming estimation method can improve dramatically the performance ofiterative CPM demodulation and decoding algorithm.Finally,some important issues of LDPC codes applicated in B3G/4G mobilesystems are discussed.A semirandom construction method of quasi-cyclic LDPC codesis proposed.With this method,a LDPC code with more flexible parameters is obtained.And the fixed-point implementation of LDPC decoding algorithm have been analyzed.Due to the rate-compatible property ofπ-rotation LDPC codes,a new adaptive LDPCscheme is designed,and its performance over fading channels is investigated.Furthermore,an iterative multiple-input multiple-output detection and LDPC decodingalgorithm in a B3G/4G TDD downlink is derived,and the performance over COST207channels are also shown.更多还原