Blind multipath identification using compressive sensing for through-the-wall noise imaging radar

In through-the-wall radar applications, multipath effects due to room reverberations result in various undesirable phenomena including target-shift, image defocusing and false alarms. Under the sparse channel assumption proved by finite difference time domain simulations, this paper proposes a blind multipath channel estimation approach by extending the blind compressive sensing dictionary learning technique to the Toeplitz structured random noise sensing matrix. The approach iteratively minimizes the estimation error of sensing matrix and channel sparsity within the structural constraints of the sensing matrix driven by random noise sequence. Computer simulations demonstrate the efficiency and capabilities of the proposed algorithm.