Soydan Redif

Signal Processing · Polynomial Matrix Decompositions · PEVD/PGEVD · Convolutive Blind Source Separation

My research concerns broadband and multichannel signal processing, with particular emphasis on polynomial matrix decompositions, polynomial eigenvalue decomposition (PEVD), polynomial generalised eigenvalue decomposition (PGEVD), para-Hermitian polynomial matrices, filter banks, and convolutive blind source separation (BSS).

This page presents selected research concepts and original explanatory figures associated with this work.

Research Topics

Polynomial matrix decompositions
Polynomial eigenvalue decomposition (PEVD)
Sequential matrix diagonalization (SMD)
Polynomial generalised eigenvalue decomposition (PGEVD)
Convolutive and broadband blind source separation
Para-Hermitian polynomial matrices
Paraunitary filter banks
Broadband multichannel signal processing

Selected Research Figures

Polynomial GEVD framework for convolutive broadband blind source separation developed by Soydan Redif. — **Polynomial GEVD for convolutive broadband blind source separation.** Polynomial generalised eigenvalue decomposition applied to non-stationary broadband blind source separation.

Read the detailed PGEVD and convolutive BSS research page →

Sequential matrix diagonalization for polynomial eigenvalue decomposition associated with Soydan Redif. — **Sequential Matrix Diagonalisation (SMD) for Polynomial Eigenvalue Decomposition.** SMD was introduced by S. Redif, S. Weiss and J. G. McWhirter in *IEEE Transactions on Signal Processing*, vol. 63, no. 1, pp. 81–89, 2015. It sequentially reduces off-diagonal polynomial energy to obtain a spectrally majorised polynomial eigenvalue decomposition of a para-Hermitian matrix.

Polynomial matrix decompositions and paraunitary filter banks associated with Soydan Redif. — **Polynomial matrix decompositions and paraunitary filter banks.** Polynomial-matrix approaches to broadband multichannel signal processing.

Research Profiles

Google Scholar ORCID