Title :
The correlation preserving transform resolves phase ambiguity in complex decorrelating transforms
Author :
Cheong Took, C. ; Douglas, Scott C. ; Mandic, Danilo P.
Author_Institution :
Dept. of Comput., Univ. of Surrey, Guildford, UK
Abstract :
A solution to the problem of intraference is obtained by recognising that the current diagonalisation schemes for the complex symmetric pseudocovariance matrix are not adequate to preserve its intrinsic complex-valued nature. To this end, we propose the correlation preserving transform (CPT), which both maintains the degree of the intrinsic correlation within a recovered bivariate source (minimum intraference) and minimises the correlation between different bivariate sources (minimum interference). Unlike the existing literature, this work therefore considers both the correlation structure between the data channels within bivariate sources and inter-source correlation. The efficacy of the proposed method is validated through analysis and simulations.
Keywords :
interference (signal); signal processing; CPT; bivariate source; complex decorrelating transforms; correlation preserving transform; correlation structure; intersource correlation; phase ambiguity; pseudocovariance matrix; signal processing; source correlation; Approximation methods; Correlation; Covariance matrices; Matrix decomposition; Signal processing; Symmetric matrices; Transforms;
Conference_Titel :
Sensor Array and Multichannel Signal Processing Workshop (SAM), 2014 IEEE 8th
Conference_Location :
A Coruna
DOI :
10.1109/SAM.2014.6882341