Conference Paper

An improved proportionate NLMS algorithm based on the l0 norm

Telecommun. Dept., Univ. Politeh. of Bucharest, Bucharest, Romania
DOI: 10.1109/ICASSP.2010.5495903 Conference: Acoustics Speech and Signal Processing (ICASSP), 2010 IEEE International Conference on
Source: IEEE Xplore

ABSTRACT The proportionate normalized least-mean-square (PNLMS) algorithm was developed in the context of network echo cancellation. It has been proven to be efficient when the echo path is sparse, which is not always the case in real-world echo cancellation. The improved PNLMS (IPNLMS) algorithm is less sensitive to the sparseness character of the echo path. This algorithm uses the l1 norm to exploit sparseness of the impulse response that needs to be identified. In this paper, we propose an IPNLMS algorithm based on the l0 norm, which represents a better measure of sparseness than the l1 norm. Simulation results prove that the proposed algorithm outperforms the original IPNLMS algorithm.

  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Acoustic echo is one of the most important issues in communication. It creates disturbance in day-to-day communication. This echo can be cancelled using adaptive filters which are governed by adaptive algorithms. Right from the introduction of Least Mean Square (LMS) algorithm, over the years, a lot of research has been done in this field in order to develop new algorithms which can effectively drive the filter to give better performance. In this review paper, we have studied and discussed all the previous work done on these algorithms in relation to acoustic echo cancellation. This paper contains the basic review of all such existing algorithms as well as their merits and demerits. It covers the basic algorithms like LMS algorithm,Recursive Least Square algorithm as well as their modified versions like Normalized Least Mean Square algorithm, Fractional Least Mean Square algorithm, Filtered-x Least Mean Square algorithm etc. Finally, a tabular comparison has been given towards the end of the paper in order to conclude the discussion.
    IEEE Sponsored International Conference on Communication and Signal Processing, India; 04/2014
  • Source
    International Journal of Applied Engineering Research. 10/2014; 9(17):3781-3805.
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Acoustic echo is one of the most important issues in full duplex communication. The original speech signal is distorted due to echo. For this adaptive filtering is used for echo suppression. In this paper our objective is to cancel out the acoustic echo in a sparse transmission channel. For this purpose many algorithms have been developed over the period of time, such as Least Mean Square (LMS), Normalized LMS (NLMS), Proportionate NLMS (PNLMS) and Improved PNLMS (IPNLMS) algorithm. Of all these algorithms we carry out a comparative analysis based on various performance parameters such as Echo Return Loss Enhancement, Mean Square Error and Normalized Projection Misalignment and find that for the sparse transmission channel all these algorithm are inefficient. Hence we propose a new algorithm modified -mu - PNLMS, which has the fastest steady state convergence and is the most stable among all the existing algorithms, this we show based on the simulation results obtained.
    Springer International Conference On Advanced Computing, Networking and Informatics, India; 06/2014

Full-text (2 Sources)

Available from
Jun 5, 2014