Estimating Symbol Duration of Long-Code Direct Sequence Spread Spectrum Signals at a Low Signal-to-Noise Ratio

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.