摘要

Frame detection is an important step in satellite-based automatic identification system for its contributions in verifying the presence of automatic identification system signal before frame synchronization. In this paper, a constant false alarm rate frame detector is proposed, which exploits the feature implied in the training sequence, to realize frame detection in the presence of additive white Gaussian noise and frequency offset. False alarm probability is related with a threshold, which is independent of the signal and noise. For fixed false alarm probability, the relationship between detection performance and E-b/N-0 is analyzed. Simulations prove that the proposed detector outperforms the detector based on cyclic autocorrelation when message collisions exist.