1: \begin{abstract}

2: 

3: 

4: This paper introduces a new algorithm to approximate non orthogonal joint diagonalization (NOJD) of a set of complex matrices. This algorithm is based on the Frobenius norm formulation of the JD problem and takes advantage from combining Givens and Shear rotations to attempt the approximate joint diagonalization (JD). It represents a non trivial generalization of the JDi (Joint Diagonalization) algorithm (Souloumiac 2009) to the complex case. The JDi is first slightly modified then generalized to the CJDi (i.e. Complex JDi) using complex to real matrix transformation.

5: Also, since several methods exist already in the literature, we propose herein a brief overview of existing NOJD algorithms then we provide an extensive comparative study to illustrate the effectiveness and stability of the CJDi w.r.t. various system parameters and application contexts. %Firstly, we have transformed complex matrices to the real onces in order to use an existing algorithm for real JD. Secondely, by observing real matrices structure, we have have developed our new algorithm (referred as CJDi Complex Joint Diagonalization) by considering the complex matrices and improved performances in terms of computational cost, convergence rate and JD quality. Numerical simulations illustrate the efficiency and stability of proposed algorithms and compared with respect to some existing  non orthogonal JD methods (ACDC, FAJD, QDiag, UWEDGE, JUST, CVFFDiag, ALS and LUCJD). 

6: 	

7: \keywords{Non orthogonal joint diagonalization, Performance comparison of NOJD algorithm, Givens and Shear rotations.}

8: \end{abstract}