Blind channel identification with colored sources by exploiting properties of companion matrices

This paper proposes a new second-order statistics-based method for blind multiple-input multiple-output (MIMO) finite-impulse-response (FIR) channel estimation driven by colored sources. It is assumed that the second-order statistics (SOS) of the input sources are known a priori. By exploiting the new derived properties of the companion matrices, an original proof for the uniqueness of the system solution is provided, which serves as a theoretical basis for the proposed new method that admits a closed-form solution. The corresponding identifiability conditions and the computational complexity of the proposed method are discussed and compared with other existing method. Numerical simulation results are presented to illustrate the performance of the proposed algorithm.