Signals Detection Of Satellite-Based AIS Based On Characteristic Sequences

Automatic identification system (AIS) is a wireless communication system. It can beused to identify and communicate each other at sea, including between ships and shore units.Conventional AIS has the limit of distance and it only can be used to supervise the key waters.Breaking this limitation, Satellite-based AIS receives the AIS terminal signals through thesatellite and feedbacks the information to monitoring center of the shore to achieve thepurpose of monitoring ships in the large range. However, influenced by the remote distanceand relative motion between the satellite and ships, received AIS signals cannot be detectedcorrectly which are with low SNR, a large range of time delay and Doppler frequency. Basedon this, this dissertation studies the signal detection from two aspects, analysis theshortcomings of current algorithms and puts forward the improvement schemes:(1) According to a wide range of time delay existing in the AIS transmission, thisdissertation analyses the characteristics of AIS frame structure and introduces a newalgorithm, for symbol synchronization, in which, interval synchronization code areconstituted by training sequences, starting flag and ending flag. Frame synchronization isobtained by differential correlation functions．Simulation results show that the algorithm isnot only insensitive to frequency shift, but also has better anti-noise performance comparedwith difference convolution at low Signal to Noise Ratio (SNR).(2) Aiming at a wide range of Doppler frequency shift and low SNR existing in the AIStransmission, this dissertation proposes a new design algorithm based on the theory ofconventional PLL. The new design is aided by ML estimating. In the algorithm, frequencyoffset is reduced to fast capture zone after compensations to the received AIS signals and thenphase locked loop starts to track surplus frequency. The simulation results show that thealgorithm has a good tracking performance under large frequency offsets and low SNRcondition.In addition, this dissertation provides and verifies the performance of the linear coherentdemodulation algorithm and the Viterbi coherent demodulation algorithm which is based onthe maximum likelihood sequences detector (MLSD). The simulation results show that bothalgorithms have better performances of demodulation.