System Identification Using Sparsity-Aware Ad...

System Identification Using Sparsity-Aware Adaptive Filtering Algorithms

Author Name : Senthil Murugan.B, Sundaravanan.J

ABSTRACT

In order to improve the performance of least mean square based adaptive filtering for identifying an unknown system, a new adaptive filtering algorithm called Sparsity-Aware algorithm is proposed in this paper. For this, an extensive literature study on various adaptive algorithms like Least Mean Square, Normalized Least Mean Square(LMS), Recursive Least Square(RLS) algorithms are made.LMS algorithm is simple to implement and has less computational complexity than NLMS and RLS. Stability has always been an issue with LMS algorithm, RLS is complex algorithm but it works more efficiently. All these algorithms works on the basis of Least Mean Square Error and filter’s weights are recursively updated as to bring output signal equal to the desired signal. There is a room for improvement in system identification using adaptive filtering in terms of computational complexity, convergence rate and misadjustment error. LMS and RLS algorithms are applied to the unknown system and the simulation results are obtained.Simulation results illustrate that the proposed algorithms outperform standard LMS and RLS in both convergence rate and steady-state performance for sparse systems.

Keywords: Least Mean Square Error(LMSE), Convergence rate, Computational Complexity, Misadjustment error.