Independent Component Analysis of Complex Valued Signals Based on First-order Statistics

Abstract

This paper proposes a novel method based on first-order statistics, aims to solve the problem of the independent component extraction of complex valued signals in instantaneous linear mixtures. Single-step and iterative algorithms are proposed and discussed under the engineering practice. Theoretical performance analysis about asymptotic interference-to-signal ratio (ISR) and probability of correct support estimation (PCE) are accomplished. Simulation examples validate the theoretic analysis, and demonstrate that the single-step algorithm is extremely effective. Moreover, the iterative algorithm is more efficient than complex FastICA under certain circumstances.