This paper presents a comprehensive Non Line of Sight (NLOS) localization scheme and a least square estimator that leverages on the bi-directional estimation of the Angle of Arrival (AOA) and Time of Arrival (TOA) of signals exchanged between mobile and reference devices. The proposed localization scheme requires two or more signal paths which can be either Line of Sight (LOS) signals or Non-line of Sight (NLOS) multipath signals that undergo one bound scattering. Our multipath selection scheme is shown to be able to discard multiple bound scattering paths with a high degree accuracy. We used empirical data obtained through experimentation in a real environment to analyse the performance of our proposed localization scheme, and to compare it to the existing methods. The results of this experiment show that the proposed localization scheme that just uses two signal paths, is able to outperform the existing localization schemes in both LOS and extreme NLOS situations where all reference devices are in NLOS with the mobile device. This localization approach is very useful in multipath environments where it may not always be possible to have at least three reference devices in LOS with the mobile device.