A parametric subspace-based blind estimation of a SIMO-FIR with unknown channel order

Subspace techniques have become a popular tool for blind identification of SIMO-FIR systems. A serious drawback of these methods is their sensitivity to the order modeling errors. We show that the so-called "parametric" approach of subspace methods (which exploits the specular structure of the propagation channel), is robust to channel order overestimation. We propose a deflation procedure to resolve the parametric subspace criterion. Simulations show that this algorithm outperforms classical subspace methods and remain consistent even when the channel order is overestimated by a large factor.