TARGET: A direct AOA-TDOA estimation for blind broadband geolocalization

In this paper, a new robust and low computationally algorithm is proposed for broadband geolocalization. Recent work have demonstrated the superiority of the geolocalization in 1-step over the 2-steps algorithms. However this superiority is obtained at the price of a bandwidth slicing which is unfortunately limited for computational reasons and leads to an asymptotic bias due to the residual broadband effect. This paper we propose an alternative approach fully exploiting the total bandwidth and consequently suppressing the slicing drawbacks. The proposed method is named TARGET and exploits the rank deficiency of a temporal shift dependent covariance matrix after a multichannel synchronization. Our analysis and simulations prove the performance advantage of proposed method over recently introduced ones.