Complex-valued ICA utilizing signal-subspace demixing for robust DOA estimation and blind signal separation

Abstract

This paper deals with direction of arrival (DOA) estimation and blind signal separation (BSS) based on independent component analysis (ICA) with robust capabilities. An efficient demixing procedure of complex-valued ICA is presented here, which combines the signal-subspace demixing procedure exploiting individual signal-subspace projection and Newton's iteration algorithm based on maximization of the approximate negentropy of non-Gaussian signal for array signal processing. It resolves the problems of order ambiguity and identifiability of traditional ICA for time-domain BSS. The proposed method could be directly applied to radar, sonar, radio surveillance, and communications systems for separating signals and estimating relative DOAs of signals. Several computer simulation examples for perturbations to the array manifold, unknown noise environments, and Rayleigh fading channel are provided to illustrate the effectiveness of the proposed method.

Robust DOA Estimation Using Independent Component Analysis

“…Therefore, it is crucially on performing some application-dependent preprocessing for using classical FastICA algorithm. However, in [15], a complex-valued ICA was proposed, which is based on signal-subspace demixing approach. Here, an efficient demixing procedure of complex-valued ICA is presented in this section, which combines the noise-subspace demixing procedure exploiting individual noise-subspace projection and Newton's iteration algorithm based on minimization of the approximate negentropy of non-Gaussian signal is used to extract the specific feature of noise component for array signal processing.…”

DTCWT-ICA based Improved DOA Estimation of Closely Spaced and Coherent Signals with Uniform Linear Array

“…The application of blind source separation using ICA based DOA estimation has been studied in [11]. The independent component analysis and forwardbackward spatial smoothing techniques is used for DOA estimation [12].…”

“…DOA estimation algorithms need to be modified when the signals are closely spaced and coherent with 2 resolution with relatively very low SNR for the performance enhancement. Most of the DOA estimation methods designed for non-coherent signals or coherent but under high SNR conditions [5][6][7][8][9][10][11][12][13][14]. The signals are closely spaced due to multipath propagation; they usually become coherent.…”

Space-Time Blind Multiuser Detection for Multiuser DS-CDMA and Oversampled Systems

“…Temporal methods enforce known signal properties on their estimates, and are commonly referred to property-restoral methods [8]. The independent component analysis (ICA) algorithms make use of the independence of different users' signals to achieve signal recovery [9,10]. ILSP and ILSE exploit the finite alphabet property to estimate signals while CM-based algorithms are effective to constant-module modulated signals [11][12][13][14].…”