Channel Estimation For Satellite-terrestrial Relay Of LEO Broadband Satellites

Relay transmission system of LEO broadband satellites has gained more attention owing to its wide coverage,easy information access and high transmission rate.The channel estimation is the premise of the correct detection and decoding.Taking advantage of the sparsity of ultra wideband channels,this thesis presents fast time-varying cascaded channel estimation algorithms in the satellite-terrestrial relay system to reduce the computational complexity.The major contributions of this work are summarized as follows.Conventional channel estimation algorithms over time-varying relay transmission have a high computational complexity.In this thesis,we propose a fast time-varying sparse cascaded channel estimation algorithm based on adaptive basis expansion model(BEM)with the known locations of the valid taps to reduce the complexity which considers the good anti-time-selectivity of the BEM.Firstly,we design a threshold by means of normalized doppler frequency shift and signal-to-noise ratio to implement an auto-adaptive BEM in the AF transmission.Furthermore,a reasonable LS/LMMSE rule is selected to estimate main channel coefficients,which utilizing the prior information of channel sparsity.In particular,the computational complexity and pilot cost are significantly decreased due to the reduced estimation of the coefficients in the sparse locations.The research results show that the proposed algorithms hold the estimation performance,reduce the parameters to be estimated and reduce the complexity in comparison with the conventional algorithms.As an extension of the previous work,considered the feature of reconstructing sparse signal for compressive sensing,a compressive sensing based BEM channel estimation algorithm is proposed to reduce the computational complexity and remove the requirement of the tap locations.Firstly,for a single OFDM symbol,the channel coefficients of the non-sparse locations are calculated by orthogonal matching pursuit which maximizes inner product between the residual errors.In addition,utilizing the channel correlation among multiple OFDM symbols,the estimation performance is further improved by structured orthogonal matching pursuit.The research results show that the proposed algorithms hold the accuracy of the estimation under the fast time-varing channel and reduce the impact of time-varing on the performance of estimation algorithms.The research results show that the proposed approaches can track the varying state of the LEO satellite channel in real time and afford a fundament for the accurate coherent demodulation of the OFDM system.