Research On Channel Estimation Algorithm Of OFDM System Based On Compressed Sensing

The OFDM(Orthogonal Frequency Division Multiplexing)technology has become an irreplaceable part of the wireless communication technology,and it is very valuable in practical application.Channel estimation is the key technology in OFDM communication systems,whether its performance is good or bad will directly affect the quality of the whole communication systems.The proposed compressed sensing algorithm improves the sparse channel estimation for OFDM systems efficiently,and reduces the pilots for channel estimation to get high spectrum efficiency.In this thesis,we have analyzed the channel estimation problem of OFDM systems based on compressed sensing.In OFDM systems,if the sparsity level of channel is known,the conventional compressed sensing algorithms,such as orthogonal matching pursuit(OMP)algorithm,compression sampling matching pursuit(CoSaMP)algorithm,subspace pursuit(SP)algorithm show good estimation performance with the reasonable parameters,and this presents that the compressed sensing algorithm performs excellent in channel estimation of OFDM systems.However,the sparsity level of channel is usually unknown in the real systems.So there is great limitation in the application of the compressed sensing algorithms with known sparsity in the channel estimation of the real OFDM systems.In order to do better channel estimation with compressed sensing,we need to research algorithms with unknown sparsity.Firstly,this thesis introduces the channel estimation algorithm by conventional sparsity adaptive matching pursuit(SAMP),but SAMP has the disadvantages of underestimation and overestimation as the characteristic of adaptive sparsity,so it brings the bad influence on the channel estimation.At the same time,SAMP needs to improve its computational complexity to pursuit better performance and this has a bad influence on the communication real time quality.This thesis proposes a new algorithm named regularized sparsity adaptive matching pursuit(RSAMP)algorithm to solve the above problems.The RSAMP does not need a priori knowledge of the channel sparsity compared with conventional compressed sensing algorithms,and searches the position of maximal difference to choose the supported elements,then regularizes these elements to improve the accuracy of supported set.Simulation results demonstrate that the RSAMP scheme has a superior performance and low computational complexity.At the same time,there is high peak to average power ratio in OFDM systems,which badly infects the work of power amplifier,and the clipping process could eliminate this bad influence,however,the pilots signal will be seriously infected by the nonlinear distortion which caused by clipping,and the performance of channel estimation will become bad.In order to solve this problem,this thesis puts forward to an iterative method.This method estimate the channel impulse response and the nonlinear distortion by compressed sensing respectively.It reduces the damage of thenonlinear distortion to the channel estimation by compensating the pilots.So this method expands the application of compressed sensing to estimate the channel impulse response in OFDM systems.