| Hybrid satellite-terrestrial OFDM communication system combines the advantages of satellite communication system and terrestrial communication system,which have large coverage,the long communication distance,small geographic restrictions and flexible networking.During wireless signals transmission via hybrid satellite-terrestrial channel,radio waves suffer from free space path loss,shadowed fading,multi-path effects and Doppler frequency shift,which will degrade the signal transmission quality.To make sure that channel equalization and coherence demodulation can be performed successfully,the key channel estimation technilogy is needed on the receiver to ensure the signal transmission quality.Since compressed sensing(CS)theory was proposed,fewer samples lower than those obtained by Nyquist rate could reconstruct a sparse signal successfully,and CS theory has been widely and deeply researched in applied mathematics and signal processing.Considering that OFDM channel experiences sparsity in time domain,CS theory can be perfectly applied in channel estimation of hybrid satellite-terrestrial OFDM communication system,which can decrease the pilot overhead significantly,increase the level of frequency resource utilization,and receive a satisfactory channel estimation property,compared with the conventional channel estimation method.This theme completely utilized the sparsity of hybrid satellite-terrestrial channel,and support the transparent radio interface based compressed sensing theory for the satellite and terrestrial sub-systems.This thesis investigated the compressed sensing based channel estimation methods,focused on the optimal pilot allocation of sparse channel estimation proposed and efficient and fast recovery algorithm to reconstruct the CIR of hybrid satellite terrestrial channel,which correspond to the process of exploration and reconstruction,and can make difference on theoretical and engineering application.The research contents can be specified as follows.Firstly,the estimation model of sparse channel in time domain in OFDM system was established,and the relation between CS theory and channel estimation was associating.Compared with the conventional channel estimation method,compressed sensing based channel estimation method adopted randomly distributed pilot pattern and orthogonal matching pursuit(OMP)algorithm was selected to exploit the sparse channel.Simulation experiments proofed the feasibility of CS based channel estimation scheme,and offered theoretical and numerical basis for following studies.Secondly,two designs for optimized pilot patterns were proposed,which can improve estimation property in aspect of exploration.Based on the conclusion that recovery matrix depends on the pilot allocation,two criteria based on minimizing modified mutual coherence were proposed.Related simulation of the first criteriaverified its limitation,and the improved criterion was proposed.Simulation experiments verified the property advantage of these two criteria,and the comparative result showed the improved one could obtain larger performance gains,thus could support the high efficient channel recovery algorithm.Thirdly,design efficient channel recovery algorithm,which can improve estimation property in aspect of reconstruction.With regards to the limit of OMP algorithm that requires a priori information about the sparsity level of the channel,sparsity adaptive matching pursuit(SAMP)algorithm was proposed.Then a new algorithm improved adaptive step size-SAMP(AS-RSAMP)and adaptive step size-regularize SAMP(AS-RSAMP)were proposed after analyzing the pros and cons of SAMP algorithm.On one hand,these two algorithms maintained the unconstrained sparsity characteristic.On the other hand,the threshold to stop iteration,the method of choosing step size and threshold to update the index were improved.Simulation results showed that AS-SAMP and AS-RSAMP had an advantage of computation speed or recovery accuracy,respectively. |
PDF Full Text Request