低信噪比下分组交织器识别

针对现有的分组交织器识别算法计算复杂高且容错性差缺点，从分组交织后的同步码分布规律出发，提出了一种新的识别算法。首先，利用数据矩阵统计特性，给出了在任意矩阵列数下，同步码和随机业务数据位置上的概率密度分布函数，基于最小错误判决准则，设定了同步码检测门限，同时基于3倍标准差准则，求解出稳健的交织周期识别门限；其次，分析了数据矩阵中每一行与每一列累积量之间的对应关系，提出了一种快速交织周期遍历方法，使得数据矩阵的构建次数大大减少；最后，总结了4个分组交织后同步码分布规律，通过遍历同步码序列，利用同步码之间的位置关系，实现交织同步位置、分组交织列与交织行参数快速识别。仿真结果表明:所提算法具有较强的低信噪比容错性，在信噪比为-6 dB条件下，参数识别率能够达到98%以上，同时与现有的算法相比，其性能提升4~10 dB且计算效率明显提高。 

Recognition of packet interleaver at low SNR

In view of the shortcomings of the existing algorithms for blind recognition of packet interleaver, which are high computational complexity and poor fault tolerance, a new recognition algorithm based on the distribution of synchronization codes after packet interleaving is proposed in this paper. Firstly, the proposed algorithm based on the statistical characteristics of data matrix gives the function of probability density distribution for synchronous code and random traffic data in any number of matrix columns, and based on the minimum error decision criterion, the detection threshold of synchronous code is set. At the same time, the detection threshold of robust interleaving period is set based on the criterion of 3 times standard deviation. Secondly, the corresponding relationship between each row and column in the data matrix is analyzed, and a fast interleaving period traversal method is proposed, which greatly reduces the number of times of data matrix construction. Finally, the four rules of distribution of synchronization codes are summarized, and by traversing the synchronization codes and utilizing the relationship of positions between synchronization codes, the parameters of synchronization positions, interleaving column and row can be identified efficiently. The simulation results show that the algorithm has a strong error tolerance at low SNR and the correct rate of parameter recognition can reach more than 98% at the SNR of -6 dB. At the same time, compared with the existing methods, its performance is improved by 4-10 dB and the calculation efficiency is significantly improved.