Sparse Adaptive Filtering Techniques for Channel Estimation » książka
Sparse Adaptive Filtering Techniques for Channel Estimation
ISBN-13: 9783659468759 / Angielski / Miękka / 2015 / 76 str.
Recently, sparse signal approximation has become an increasingly important research area in signal processing. It attracts a lot of interest due to its wide range of practical applications. In this work, a novel adaptive filtering algorithm with relative low computational complexity that is capable of exploiting the sparsity of systems is proposed. The basic idea here is, we adopt a p-norm constraint in the cost function of the variable step-size least mean square (VSSLMS) algorithm. This constrain imposes a zero attraction at each filter coefficient based on their respective relative value. Also, the convergence analysis of the proposed algorithm is presented and the stability condition is derived. The performance of the proposed algorithm has been compared to those of the Zero Attraction Least Mean Square(ZA-LMS), windowing ZA-LMS(wZA-LMS), Non-uniform Norm Constraint LMS(NNCLMS) in a system identification setting for different additive Gaussian noise(AGN), additive correlated noise(ACN)and additive impulsive noise(AIN) environments. The proposed algorithm has always shown superior performance to the others with less or comparable number of computations
Recently,sparse signal approximation has become an increasingly important research area in signal processing. It attracts a lot of interest due to its wide range of practical applications. In this work, a novel adaptive filtering algorithm with relative low computational complexity that is capable of exploiting the sparsity of systems is proposed. The basic idea here is, we adopt a p-norm constraint in the cost function of the variable step-size least mean square (VSSLMS) algorithm. This constrain imposes a zero attraction at each filter coefficient based on their respective relative value. Also, the convergence analysis of the proposed algorithm is presented and the stability condition is derived. The performance of the proposed algorithm has been compared to those of the Zero Attraction Least Mean Square(ZA-LMS), windowing ZA-LMS(wZA-LMS), Non-uniform Norm Constraint LMS(NNCLMS) in a system identification setting for different additive Gaussian noise(AGN), additive correlated noise(ACN)and additive impulsive noise(AIN) environments. The proposed algorithm has always shown superior performance to the others with less or comparable number of computations.
Czytaj nas na:
KrainaKsiazek.PL - Księgarnia Internetowa
