A Low-Complexity Sparse LMS Algorithm Optimized for Hardware Implementation.

A novel sparse LMS algorithm for the sparse system identification is proposed to reduce the complexity of the existing algorithms. The proposed algorithm discards W(n) when the value of the current weight vector is within a certain range. The algorithm also optimizes the iterative update equation by using only the product term to calculate the value of W(n+1). The hardware implementation shows that the total logic elements number of the proposed algorithm is 60.06% less than the latest l0-ILMS algorithm, while the performance is nearly the same. [ABSTRACT FROM AUTHOR]