This paper presents an online estimation scheme for the identification of distribution power system impedance parameters. The proposed scheme is based on an invasive current injection method. The scheme benefits from a fast and accurate adoptive notch filter for the extraction of the ambient voltages and currents and also a frequency domain least squares estimator. The former enhances the scheme by providing accurate information even under corrupted grid signals, while the latter purifies the estimations from the measurements noise. Simulation and experimental results verify the accuracy of the proposed scheme and its ability to perfectly track the impedance variations of the distribution network.