Studies On Adaptive Sparse Channels Estimation Algorithms

Compared with wired communication system,the transmission environment of wireless communication system is more complex,so its performance will be affected by wireless transmission channel.In particular,the multipath effect caused by multipath delay expansion in wireless communication channel leads to serious intersymbol interference of the received signals and affects the quality of wireless communication.In order to improve the quality of wireless communication,reduce the bit error rate of the system and eliminate the intersymbol interference caused by multipath effect,it is necessary to implement an effective estimation of wireless communication channel,which is called as channel estimation.In addition,the impulse response of wireless communication channel is always sparse due to the effect of multipath effect.Therefore,this paper aims to analysis the inherent sparse characteristics of wireless channels to research and develop sparse channel estimation algorithms achieving effective estimation for wireless sparse channel.In the process of research,the developed algorithms can improve the performance of the existing channel estimation algorithms,which are verified by simulation experiments.The results show that the proposed channel estimation algorithms are effective.In this paper,the detail research works are as follows:For the low signal-to-noise ratio(SNR)communication environment,a zero-attracting l2-lp algorithm is developed to realize the design of low signal-to-noise sparse adaptive channel estimation algorithm,which is obtained by using mixed error,compressed sensing and l1-norm.In order to solve the problem that the zero-attracting l2-lp algorithm has the same zero-attracting ability for all of the coefficients in the sparse channel,sparse l2-lp algorithms are developed by utilizing logarithm weighted and correntropy induced metric criterion.After that,the developed algorithms are analyzed by approximate method,and the constraint boundary condition of the step size is obtained.At last,the developed new algorithms are verified by simulating and comparing under low SNR environment.The simulation results show that the proposed algorithms can effectively reduce the steady-state estimation error and improve the convergence speed for sparse channel estimation.For Gaussian environment,the low-complexity sparse adaptive channel estimation algorithms(zero-attracting SM-NLMS(Set-membership normalized least mean square)algorithm,reweighting zero-attracting SM-NLMS algorithm,correntropy induced metric constrained SM-NLMS algorithm)are developed by using l1-norm,logarithm weighted and correntropy induced metric criterion based on compressed sensing and set-membership theory,respectively.For further reduce the computational complexity and steady-state estimation error,a linear function approximation method is proposed,then a linear function approximation constrained SM-NLMS(LFASM-NLMS)algorithm is developed.By analyzing the performance of the proposed zero attracting items and computational complexity of the proposed new sparse SM-NLMS algorithms,it is found that the developed LFASM-NLMS algorithm has relatively low computational complexity and superior steady-state error performance.Hence,based on the principle of energy conservation,the steady-state convergence analysis of the proposed LFASM-NLMS algorithm is carried out by using the approximate method,and the steady-state mean square error of the proposed LFASM-NLMS algorithm is obtained.The performance of the proposed new algorithms is verified by estimating time-varying sparse channel under Gaussian environment.For non-Gaussian environment and sparse channel,a robust sparse maximum correntorpy criterion algorithm(mixed norm constrained maximum correntorpy criterion algorithm,MN-MCC)is developed by using MCC criterion and mixed norm penality.After that,the proportional sparse maximum correntropy criterion algorithm is developed by using gain distribution matrix.At last,the performance and robustness of the proposed new algorithms are verified by simulating under different input signals(white input signal and colored input signal)and noise signals(mixed Gaussian noise and Alpha-stable noise).The simulation results show that the proposed new algorithms have lower steady-state estimation error and faster convergence speed.At last,for a typical sparse underwater acoustic communication channel,an underwater acoustic channel estimation system is proposed,and the real underwater acoustic channel in the system is simulated by using Waymark.According to the above different environment,the proposed algorithms are verified by estimating Waymark modelling underwater acoustic channel in the proposed underwater acoustic channel estimation system.The simulation results show that the proposed algorithms can effectively estimation the real underwater acoustic channel.