ICA with time-varying convergence factor and its application in communications
Thomas T. Yang, Wasfy B. Mikhael · International Conference on Communications · 2006
Blind signal separation in dynamic environment is a challenging problem. This paper presents a new gradient-based Independent Component Analysis (ICA) technique for dynamic situations. The algorithm optimizes the convergence factor for each iteration during the adaptation process. Compared with previously proposed ICA with Optimum Block Adaptation (OBA/ICA), the new technique eliminates the matrix inversion operation and is applicable in the cases where the block size is less than the dimensionality of the system. Computer simulation results for interference suppression in wireless communications confirm the effectiveness of the proposed technique. It compares favorably with the well-known fast-ICA and online gradient ICA algorithms.