This paper address the problem of direction-ofarrival (DOA) estimation of quasi-stationary signals based on uniform linear array with malfunctioning sensors. By utilizing the subspace structures of the local second-order statistics of quasistationary signals, a Khatri-Rao subspace approach is developed. Our scheme first collects the local covariance matrices of the source signals and then transfers them into a new virtual linear array which can identify at least twice as much DOAs as to the original physical one. It is also shown that the coprime configuration is a special case of the proposed model therefore the same techniques can be applied directly. Simulations are also carried out for the comparison of the proposed algorithm and state-of-the-art approaches.

parameter estimation; direction-of-arrival estimation; malfunctioning sensors; coprime array