Robustness of blind subspace based techniques using ℓp quasi-norms

Abstract

It has been very recently noted that it is possible to recover the blind channel estimate in case of channel order overmodeling by using ℓp quasi norms. But, to the best of our knowledge, there is, until now, no theoretical results that investigate this issue. In this paper, we propose to study the robustness of subspace blind methods using ℓp quasi-norms in the noiseless case and for nonsparse channels. More particularly, we provide conditions that ensures channel identifiability and study their frequency of occurence with respect to the system parameters.