Title :
A batch processing constant modulus algorithm
Author :
Xu, Changjiang ; Li, Jian
Author_Institution :
Dept. of Electr. & Comput. Eng., Univ. of Florida, Gainesville, FL, USA
Abstract :
We present a batch processing constant modulus algorithm (BP-CMA) derived by a nonlinear optimization approach to minimizing the constant modulus (CM) criterion. BP-CMA is a line search iteration algorithm. The search direction may be taken as deepest descent direction or Newton direction. The exact step size is obtained from the roots of a cubic equation. The initial value is calculated by using the eigenvectors of the signal subspace. The BP-CMA with the Newton direction has a fast convergence rate and can converge to the minima of the CM criterion after a few iterations.
Keywords :
blind equalisers; blind source separation; convergence; eigenvalues and eigenfunctions; iterative methods; minimisation; Newton direction; batch processing constant modulus algorithm; cubic equation; eigenvectors; fast convergence rate; line search iteration algorithm; minimisation; nonlinear optimization; signal subspace; step size; Blind equalizers; Blind source separation; Convergence; Covariance matrix; Data models; Nonlinear equations; Signal processing; Stochastic processes; Symmetric matrices; Vectors; Batch processing algorithm; constant modulus algorithm; signal-subspace;
Journal_Title :
Communications Letters, IEEE
DOI :
10.1109/LCOMM.2004.835330
