【作者】 殷玮玮；

【导师】 吴乐南；

【作者基本信息】 东南大学 ， 信号与信息处理， 2006， 博士

【摘要】 无线通信和多媒体通信技术的发展改变了人类的生活,下一代无线移动通信的目标是实现无所不在的,高速率、高质量的多媒体移动传输。然而无线信道中有限的带宽、恶劣的信道条件以及各种终端尤其是便携式移动设备对复杂度、能耗等资源的苛刻限制,使传统依据Shannon分离定理将信源和信道作为两个独立模块分别设计的思想逐渐显露出了局限性。实际系统中一方面由于编码复杂度、延时的限制,同时为了提高系统的鲁棒性,信源编码输出中仍留下一定的冗余;另一方面由于带宽有限和无线信道中的衰落、噪声等因素使信道编码无法提供充足的保护,传输中的差错不可避免。因此突破传统常规的系统设计方法,将信源编码和信道编码之间进行联合信息处理以寻求更大的性能增益是一个必然趋势。本论文研究的信源-信道联合编码/调制就是将信源编/译码和信道编/译码及调制综合考虑、设计和优化的一种策略,从整体上优化通信系统,减小包含信源和信道的整个系统的端到端错误和失真。主要获得以下几方面的成果:1、提出了一种信源-信道联合编码调制的方案。通过置换搜索算法优化信源符号和调制星座点间的映射关系以减小信道错误造成的信源均方误差失真。同时信源可以分配更多比特进行编码,减小了量化失真。这种低复杂度信源-信道联合编码调制系统的端到端性能超过了同样数据传输率下中等复杂度的级联编码系统,并且对于信道信噪比的波动具有更好的鲁棒性。2、研究了利用图像压缩编码经过噪声信道传输后获得的软输出进行信源译码重建的问题。由SOVA算法对接收序列中的信源残留冗余进行利用,不仅减小了符号错误率,同时获得了符号值的软信息。由于信道中传输的图像压缩域符号值的误差平方和与重建图像的PSNR存在对应关系,利用符号软信息对传输的压缩域符号值进行最小均方误差估计,并以此重建图像,提高了PSNR。3、针对采用训练序列获得信源中残留冗余结构的不精确性及其传输代价,提出一种基于隐马尔可夫模型,在接收端利用信道译码的软输出对信源残留冗余概率结构进行参数估计,并且和信源-信道迭代联合译码结合,在完成参数估计的同时实现了联合译码。同时提出利用一种鉴别信息度量参数精度和伴随迭代的改善程度,并用其设定停止估计迭代的条件。4、提出一种将高维的压缩图像残留冗余统计模型分解以获得低复杂度信源-信道联合译码的方法。为避免变长码的错误传播,我们往往留下这部分冗余信息在接收端加以利用以实现信源-信道联合译码,而图像压缩编码后的残留冗余仍然具有二维马尔可夫随机场模型表征的统计特性。由于直接利用场模型的联合译码算法的巨大复杂度,本文首先将二维的场模型分解为4个独立的一维统计相关性,并根据这种分解后的统计模型,分别提出了基于串行和基于并行的两种不同的低复杂度联合译码方法。更多还原

【Abstract】 The development of wireless and multimedia communications has changed people’s lives dramatically. The next generation wireless communications call for ubiquitous, high rate and high quality mobile multimedia transmission. However, limited bandwidth, severe channel and the restriction of resources such as the complexity and power in the terminal especially portable equipments, reveal the limitation of the independent design of source and channel based on Shannon’s separation theorem.On the one hand, some residual redundancy is inevitable in the output of the source encoder because of the constraint of complexity and delay in practical encoders, and some redundancy is left in order to improve the robustness at the same time; on the other hand, perfect protection can not be provided by channel encoder because of the limited bandwidth and the effect of the fading and noise simultaneously in wireless channels, so transmission errors are inevitable. Therefore, the novel system design scheme which breaks away from convention and joint information processing to seek for greater performance improvement is needed.Joint source-channel coding/modulation combined source coding (decoding) with channel coding (decoding) and modulation to optimize the communication system synthetically. The destination of this technique is to optimize the communication system from the whole and to reduce the error and distortion from point to point in the system including source and channel. The mainly work is described as follows:1、A joint source-channel coding and modulation scheme is presented. The mapping between source symbols and modulation signals is optimized through the switch search algorithm to decrease the mean square distortion of the false source symbols induced by channel errors. At the same time, the source quantizing distortion is reduced because more bits are allocated to source encoding. The end-to-end performance and robustness of the proposed low complexity joint source-channel coding and modulation system exceed the tandem system.2、The reconstruction of compressed images from the soft output of noisy channel is studied. Combined with SOVA algorithm, the redundancy between the symbols after compression is utilized. Not only the symbol error rate is reduced, but also the soft information of symbols is obtained. According to the correlation between the error square of the symbol value in compression field and the PSNR of reconstructed image, the symbol value is estimated by MMSE estimation with the soft information and used to reconstruct the image to improve PSNR.3、In order to avoid the inaccuracy and transmission cost of the residual redundancy information obtained by the training sequence, an algorithm to estimate the statistical structure of更多还原