Using Partial Zero Subspace in Deterministic Blind Channel Identification

A new blind channel identification method is proposed in this work. In single-input multi-output channel system, the outputs of subchannels are related to each other due to the fact that all subchannels are driven by an identical source. The famous least- squares approach exploits such cross relation to identify channels. It is deterministic and naturally able to identify channels driven by color signal. We study the least-squares approach and find a simplifying method by using partial zero subspaces. The new approach is more computationally efficient, because the number of equations to be solved is much fewer. Correctness of the proposed method is proved as well. As it shows, the simplification is not straightforward and is nontrivial. Simulation results are used for illustrating the performance.