A Modified Algorithm for Blind Detection of the Burst DSSS Signal

To detect the burst Direct Sequence Spread Spectrum (DSSS) signal blindly, a modified algorithm based on the feature derived from cycle spectrum is proposed. The algorithm mainly studies the construction of detection feature and the determination of the start and end points of signal. Firstly, the variance of cycle frequency α=0 profile data and the mean of variance of several profiles data close to cycle frequency α=fc are calculated respectively, and then provides the absolute value of difference between them as the detection feature. Then, the change of signal and noise is tracked by setting corresponding feature collection in real time, which can adjust the thresholds used for the decision of the start and end points of signal self-adaptively, and the points can be captured accurately. Ultimately, the correct detection of the signal is achieved. The simulation results demonstrate that the algorithm has better robustness and higher precision of decision, and the probability of detection can reach above 90% when the signal to noise ratio (SNR) is -5dB.