Research On Compressive Sensing Theories And Their Sparse Cognitive Wireles Channel Estiation

In wireless communication systems,there are interferences such as multipath fading in signal transmission,which affect the reliable reception of signals and need to be compensated by channel estimation.However,traditional and modern sparse channel estimation algorithms all fail to take full advantage of frequency-domain sparseness of wireless channels,resulting in poor performance,low accuracy and high complexity.To solve these problems,this dissertation mainly studies the Compressed Sensing(CS)theory,and it uses the theory to improve the sparse channel estimation to achieve accurate estimation.First,the overview of channel estimation,key issues and the application background of CS theory are briefly introduced.Then,the sparse wireless channel estimation model with the CS theory are established.Next,aiming at the shortcomings of the existing CS sparse reconstruction algorithms,such as low accuracy and high complexity,three reconstruction algorithms are proposed and applied to sparse channel estimation,and their simulation and analysis verify their effective improvement in accuracy and complexity.Finally,this kind of algorithms is summarized and prospected.The contents and innovations are as follows:1.To overcome the shortcomings of existing channel estimation that can not make full use of channel sparsity,CS theory is applied to channel estimation.By reconstructing signals to estimate the channel impulse response,Homotopy method is proposed,which improves the accuracy.The Homotopy idea is introduced into the algorithm,and it estimates the step and direction of the signal's path change continuously according to the previous estimates,then the final estimation is obtained.This algorithm makes full use of the sparse characteristics of the channel itself and obtains more effective and accurate estimation.The simulation results show that the algorithm has higher reconstruction accuracy and faster convergence speed.Compared with the traditional Least Squares(LS)channel estimation,the estimation accuracy is improved.Under the same reconstruction mean square error,the signal-to-noise ratio gain is nearly 20 dB compared with LS algorithm.Meanwhile,the mean square error of signal reconstruction is further reduced by increasing the number of iterations.2.Based on the existing Generalized Orthogonal Matching Pursuit(GOMP)algorithm,an improved GOMP algorithm for stopping the iteration sparsity in time is proposed to overcome the insufficiency that the number of elements in the final atomic index support set is exactly equal to the signal sparsity.The number of atoms selected for each iteration is set appropriately to improve the estimation accuracy.At the same time,by changing the number of observed values of reconstructed signals and setting the upper limit of recovery residual between the restored signals and the original signals,a more precise range of bounded isometric constants can be obtained to improve the reconstructed success rate.The simulation results show that the proposed algorithm reduces the computational time by three quarters compared with OMP algorithm,increases the probability of successful reconstruction by 50%compared with GOMP,and reduces the mean square error of reconstruction by about 15 dB.Meanwhile,the design of appropriate signal sparsity and the number of atoms selected in each iteration can make the reconstruction result more accurate and effective,thus improving the effectiveness and accuracy of the algorithm.3.Because the original GOMP reconstructed signal only get the unique atomic index support set,which can't avoid the defect of choosing the wrong atom,and the channel sparsity can not be known in advance in practice,an improved multi-path sparsity adaptive GOMP algorithm is proposed.When reconstructing the original signals,multi-path searching for atomic index is used,and the original signal can be reconstructed without predicting the sparsity of the signal.The simulation results show that the reconstruction accuracy of the proposed multi-path sparsity adaptive GOMP algorithm is about twice as high as that of the existing GOMP algorithm,and the probability of successful reconstruction is also increased about three times.In conclusion,aiming at the sparse characteristics of cognitive radio channel,Homotopy,improved GOMP and improved multi-path sparsity adaptive GOMP algorithm are used to estimate channel impulse response.The proposed algorithms not only improve the accuracy of estimation results,but also reduce the computational complexity and save computing time,thus improving communication performance.Therefore,the proposed algorithms can be used for channel estimation in future cognitive wireless communication systems such as high performance 5G.