Several existing spreading sequence estimation algorithms of long code direct-sequence spread spectrum (LC-DSSS) signals require prior knowledge of the symbol duration, but research on symbol duration estimation techniques of LC-DSSS signals are rare currently. In this paper, we proposed a method of estimating symbol duration for LC-DSSS signals. On the basis of the missing data model, a set of sample covariance matrices are constructed from the received signal with a set of window durations. Subsequently, the diagonals that contain noise component are removed from the sample covariance matrices to eliminate the effects of the noise and then squared Frobenius norm is performed on the sample covariance matrices to eliminate the effects of the long code. After analyzing the second-order statistical characteristic of the squared Frobenius norm of the sample covariance matrices, a symbol duration estimator of LC-DSSS signals is derived. Numerical experiments demonstrate that the proposed estimator provides satisfactory estimation performance of the symbol duration for LC-DSSS signals at low signal-to-noise ratio, even in multiple access interference scenario and multipath fading scenario. Compared to the existing estimators, the proposed estimator exhibits superior performance.

现有的长码直接序列扩频（LC-DSSS）信号的几种扩频序列估计算法需要对码元持续时间有先验知识，但是目前对LC-DSSS信号的码元持续时间估计技术的研究很少。在本文中，我们提出了一种估计LC-DSSS信号的符号持续时间的方法。根据丢失的数据模型，从接收的信号中以一组窗口持续时间构造一组样本协方差矩阵。随后，将包含噪声分量的对角线从样本协方差矩阵中删除以消除噪声的影响，然后对样本协方差矩阵执行平方Frobenius范数以消除长码的影响。在分析了样本协方差矩阵的平方Frobenius范数的二阶统计特性之后，得出了LC-DSSS信号的符号持续时间估计量。数值实验表明，即使在多址干扰情形和多径衰落情形下，所提出的估计器也能以低信噪比为LC-DSSS信号提供令人满意的符号持续时间估计性能。与现有的估计器相比，所提出的估计器具有更好的性能。即使在多址干扰情况和多径衰落情况下也是如此。与现有的估计器相比，所提出的估计器具有更好的性能。即使在多址干扰情况和多径衰落情况下也是如此。与现有的估计器相比，所提出的估计器具有更好的性能。