New Adaptive Algorithms for the Rapid Identification of Sparse Impulse Responses
R. Mariane · Signal Processing · 2010
In this chapter we presented a family of algorithms developed in the last years for improving the convergence of adaptive filters when modeling sparse impulse responses. The performances of the described techniques, known as proportionate-type LMS algorithms, were illustrated through computer simulations in the identification of the digital network channels of ITU-T recommendation G.168. The first proposed approach, the PNLMS algorithm, was shown to produce fast initial convergence for sparse impulse responses, followed by a significant reduction after the fast initial period. Also, its performance was poor for non-sparse impulse responses. Improved versions of the PNLMS algorithm were then described and