Research On Message Passing Algorithms For Rapid Carrier Recovery Technology

Short burst communication is widely used in the fields of air communication and ground mobile communication due to its strong anti-interference ability and high communication quality.In communication systems,phase offset is brought in by transceivers' oscillators and relative movements between communicating devices,therefore carrier recovery technology with fast convergence speed and desirable performance in low signal-to-noise ratio must be adopted to estimate and compensate the unknown phase offset,and thus realize correct demodulation.Furthermore,in short burst digital transmission,every burst frame has extremely short time duration and adjacent frames have independent channel statistics,which brings great difficulty to carrier recovery.In order to satisfy the requirements of short burst communications,this thesis researches on the problem of rapid carrier phase recovery.Firstly,we consider the scenario where Doppler frequency offset and the duration of each frame are relatively small.In this case,Doppler frequency does not need separate estimation.To solve the problem of poor estimation accuracy of carrier phase recovery in low signal-to-noise ratio,this thesis designs message passing algorithms for joint decoding and detection based on factor graph,and achieves rapid carrier recovery of short burst communication systems.The proposed algorithm approximates the phase messages to the sums of several Tikhonov distributions to eliminate the difficulty of integration of continuous variable introduced by sum-product algorithm.In addition,Kullback-Leibler divergence minimization method is employed to avoid the explosive increase in the number of Tikhonov distributions of the passing messages in both forward and backward directions.Simulation results show that compared to the typical Discretizing Phase algorithm,the complexity of the proposed one is decreased by one third while maintaining slightly better performance.Secondly,we consider the scenario where Doppler frequency offset and the duration of each frame are relatively large.In this case,Doppler frequency does need separate estimation.Therefore,phase offset consists of phase noise and Doppler frequency offset.This thesis puts forward an improved joint decoding and detection algorithm.For purpose of computing the double integral by sum-product algorithm,this thesis discretize the frequency offset and for each discrete point,messages of phasenoise are still represented by several Tikhonov distributions.Furthermore,a scheme for updating the positions of frequency discrete points is presented.Simulation results show that the proposed algorithm gains good performance with bit error rate below 10-4 in low normalized signal-to-noise ratio from 3dB to 2.5dB.Finally,hardware design of the proposed message passing algorithms based on Tikhonov-Sum approximation is unified and discussed.